mirror of
https://github.com/langbot-app/LangBot.git
synced 2026-06-04 12:56:02 +00:00
96b041846d
* feat: add mcp and skills
* feat: add filter
* feat: modify frontend
* feat(box): add sandbox_exec tool loop for local-agent calculations
* feat(box): add host workspace mounting and sandbox_exec guidance
* feat(box): add BoxProfile with resource limits and improved output truncation
- Implement head+tail output truncation (60/40 split) so LLM sees both
beginning and final results; add streaming byte-limited reads in backend
to prevent unbounded memory usage (_MAX_RAW_OUTPUT_BYTES = 1MB)
- Define BoxProfile model with locked fields and max_timeout_sec clamping
- Add four built-in profiles: default, offline_readonly, network_basic,
network_extended with differentiated resource and security constraints
- Add resource limit fields to BoxSpec (cpus, memory_mb, pids_limit,
read_only_rootfs) and pass corresponding container CLI flags
(--cpus, --memory, --pids-limit, --read-only, --tmpfs)
- Profile loaded from config (box.profile), applied in service layer
before BoxSpec validation; locked fields cannot be overridden by
tool-call parameters
* feat(box): add obs
* refactor(box): unify box service lifecycle and local runtime
management
* refactor(box): remove legacy in-process runtime code and clean up smells
After the architecture settled on always using an independent Box Runtime
service, several pieces of compatibility code and design shortcuts were
left behind. This commit cleans them up:
- Remove `LocalBoxRuntimeClient` and `create_box_runtime_client` from
production code (moved to test-only helper).
- Remove unused `_clip_bytes` method from backend.
- Remove `__langbot_session_placeholder__` hack by making `BoxSpec.cmd`
default to empty and validating non-empty only in `runtime.execute()`.
- Extract `get_box_config()` helper to eliminate 5× duplicated config
access boilerplate.
- Remove `session_id`/`host_path`/`host_path_mode` from the LLM-facing
tool schema to enforce request-scoped session isolation.
- Fix dual shutdown path: `NativeToolLoader.shutdown()` no longer calls
`box_service.shutdown()` (handled by `Application.dispose()`).
- Simplify `_assert_session_compatible` with a loop.
- Inline client creation in `BoxRuntimeConnector`.
- Remove redundant `BOX__RUNTIME_URL` env var from docker-compose
(auto-detected by code).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add test
* fix: fix box intergration test
* feat(box/mcp): integrate MCP stdio with Box sandbox — auto-isolation, dep install, security
## Summary
When Podman/Docker is available, all stdio-mode MCP servers now automatically
run inside Box containers with dependency installation, path rewriting, and
lifecycle management. When no container runtime exists, LangBot starts normally
and stdio MCP falls back to host-direct execution.
## What changed
### MCP stdio → Box integration (mcp.py)
- Add `MCPServerBoxConfig` pydantic model for structured box configuration
with validation and defaults (network, host_path_mode, timeouts, resources)
- Auto-infer `host_path` from command/args with venv detection: recognizes
`.venv/bin/python` patterns and walks up to the project root
- Rewrite host paths to container `/workspace` paths transparently
- Replace venv python commands with container-native `python`
- Auto-detect `pyproject.toml`/`setup.py`/`requirements.txt` and run
`pip install` inside the container before starting the MCP server
- Copy project to `/tmp` before install to handle read-only mounts
- Add retry with exponential backoff (3 retries, 2s/4s/8s delays)
- Add Box managed process health monitoring (poll every 5s)
- Fix session leak: `_cleanup_box_stdio_session()` now runs in `finally`
block of `_lifecycle_loop`, covering all exit paths
- Fix retry logic: `_ready_event` is only set after all retries exhaust
or on success, not on first failure
- Enhance `get_runtime_info_dict()` with `box_session_id` and `box_enabled`
### Box security (security.py — new)
- `validate_sandbox_security()` blocks dangerous host paths:
`/etc`, `/proc`, `/sys`, `/dev`, `/root`, `/boot`, `/run`,
docker.sock, podman socket
- Called at the start of `CLISandboxBackend.start_session()`
### Box models (models.py)
- Add `BoxHostMountMode.NONE` — skips volume mount entirely
- Adjust `validate_host_mount_consistency` to allow arbitrary workdir
when `host_path_mode=NONE`
### Box backend (backend.py)
- Add `validate_sandbox_security()` call in `start_session()`
- Add `langbot.box.config_hash` label on containers for drift detection
- Handle `BoxHostMountMode.NONE` — skip `-v` mount arg
- Add `cleanup_orphaned_containers()` to base class (no-op default) and
CLI implementation (single batched `rm -f` command)
### Box runtime (runtime.py)
- Call `cleanup_orphaned_containers()` during `initialize()` to remove
lingering containers from previous runs
### Box service (service.py)
- Graceful degradation: `initialize()` catches runtime errors and sets
`available=False` instead of crashing LangBot startup
- Add `available` property and guard on `execute_sandbox_tool()`
- Add `skip_host_mount_validation` parameter to `build_spec()` and
`create_session()` — MCP paths are admin-configured and trusted,
bypassing `allowed_host_mount_roots` restrictions meant for
LLM-generated sandbox_exec commands
### Default behavior
- stdio MCP servers automatically use Box when `box_service.available`
is True (Podman/Docker detected); no explicit `box` config needed
- When no container runtime exists, falls back to host-direct stdio
- MCP Box defaults: `network=on` (for pip install), `read_only_rootfs=false`
(for site-packages), `host_path_mode=ro`, `startup_timeout=120s`
### Tests
- `test_box_security.py`: blocked paths, safe paths, subpath rejection
- `test_mcp_box_integration.py`: config model, path rewriting, venv
unwrap, host_path inference, payload building, runtime info, box
availability check
- `test_box_service.py`: `BoxHostMountMode.NONE` validation tests
* feat(box/mcp): instance-based orphan cleanup, error classification, session API, and integration tests
## Changes
### Precise orphan container cleanup
- Runtime generates a unique instance_id on startup
- Every container gets a `langbot.box.instance_id` label
- `cleanup_orphaned_containers()` only removes containers from
previous instances, preserving containers owned by the current one
- Containers from older versions (no label) are also cleaned up
- `cleanup_orphaned_containers` added to `BaseSandboxBackend` as
a no-op default method, removing hasattr duck-typing
### Fine-grained MCP error classification
- New `MCPSessionErrorPhase` enum with 7 phases: session_create,
dep_install, process_start, relay_connect, mcp_init, runtime,
tool_call
- Each phase in `_init_box_stdio_server()` sets the error phase
before re-raising, enabling precise failure diagnosis
- `retry_count` tracked across retry attempts
- `get_runtime_info_dict()` exposes `error_phase` and `retry_count`
### GET /v1/sessions/{id} API
- `BoxRuntime.get_session()` returns session details including
managed process info when present
- `handle_get_session` HTTP handler + route in server.py
- `BoxRuntimeClient.get_session()` abstract method + remote impl
### stdio defaults to Box when runtime is available
- `_uses_box_stdio()` checks `box_service.available` instead of
requiring explicit `box` key in server_config
- `BoxService.initialize()` catches runtime errors gracefully,
sets `available=False` instead of crashing LangBot startup
- When no container runtime exists, stdio MCP falls back to
host-direct execution
### Code quality (from /simplify review)
- Extracted `_VENV_DIRS` / `_VENV_BIN_DIRS` module-level constants
- Removed dead `_box_network_mode()` method and unused `bc` variable
- Fixed broken import `from ....box.models` → `from ...box.models`
- Cached `_resolve_host_path()` result — computed once, passed through
- Config hash now includes `host_path` field
- Batched orphan cleanup into single `rm -f` command
### Session leak fix
- `_cleanup_box_stdio_session()` now runs in `_lifecycle_loop`'s
finally block, covering all exit paths (normal shutdown, error,
retry, final failure)
### Integration tests
- 6 end-to-end tests covering managed process lifecycle, WebSocket
stdio bidirectional IO, session cleanup verification, single
session query, process exit detection, and orphan cleanup safety
* refactor: use rpc
* fix: import
* refactor(box): clean up sandbox subsystem code quality and efficiency
- Fix O(n²) stderr trimming in runtime.py with running length tracker
- Remove dead code: RESERVED_CONTAINER_PATHS, _subprocess_wait_task,
unused config_hash computation, unused imports
- Deduplicate connection callback in BoxRuntimeConnector, parse URL once
- Use enum comparison instead of stringly-typed spec.network.value check
- Replace manual _result_to_dict/_session_to_dict with model_dump()
- Cache NativeToolLoader tool definition and sandbox system guidance
- Extract _is_path_under() helper to eliminate duplicated path checks
- Import SANDBOX_EXEC_TOOL_NAME from native.py instead of redefining
- Add JSON startswith guard in logging_utils to skip futile json.loads
- Fix ruff lint errors (F401 unused imports, F841 unused variables)
* fix: ruff
* refactor(sandbox): keep box logic out of pipeline and localagent
- Move sandbox system-prompt guidance from LocalAgentRunner into
BoxService.get_system_guidance() so all box domain knowledge stays
in the box module.
- Remove standalone logging_utils.py; merge format_result_log() into
MessageHandler base class alongside cut_str().
- Strip sandbox-specific JSON parsing from log formatting; tool
results now use generic truncation.
- Revert TYPE_CHECKING changes in stage.py and runner.py that were
unrelated to this feature.
- Skip two test files affected by a pre-existing circular import
(runner ↔ app) until the import cycle is resolved in a separate PR.
* fix: ruff
* refactor(box): move box runtime to langbot-plugin-sdk
Extract self-contained box runtime modules (actions, backend, client,
errors, models, runtime, security, server) to langbot-plugin-sdk and
update all imports to use `langbot_plugin.box.*`. Keep only service
and
connector in LangBot core as they depend on the Application context.
- Update docker-compose to use `langbot_plugin.box.server` entry
point
- Update pyproject.toml to use local SDK via `tool.uv.sources`
- Remove migrated source files and their unit/integration tests
- Update remaining test imports to match new module paths
* fix: ruff
* feat: enhance sandbox api
* refactor(box): derive paths from shared host root
* fix(box): tighten sandbox exposure and restore box integration coverage
* refactor(types): remove quoted annotations under postponed evaluation
* feat(box): unify native agent tools around exec/read/write/edit
* chore(sandbox): move MCP loader changes to follow-up branch
* feat(box): add session workspace quota enforcement and SDK quota metadata
* feat(skills): add Agent Skills management system (#1917)
* feat(skills): add Agent Skills management system
Implement comprehensive skills management feature inspired by agentskills spec:
Backend:
- Add Skill and SkillPipelineBinding database entities
- Add database migration (dbm018) for skills tables
- Implement SkillManager for skill loading, matching, and resolution
- Implement SkillService for CRUD operations
- Add skills API endpoints for skill and pipeline binding management
- Integrate skill index injection into pipeline preprocessor
- Add skill activation detection in LocalAgentRunner
Frontend:
- Add Skills page with listing, search, and type filter
- Add SkillDetailDialog for create/edit with preview
- Add SkillCard and SkillForm components
- Add skills API methods to BackendClient
- Add skills entry to sidebar navigation
- Add i18n translations (en-US, zh-Hans)
Features:
- Support skill and workflow types
- Sub-skill composition via {{INVOKE_SKILL: name}} syntax
- Progressive disclosure (index in prompt, full instructions on activation)
- Pipeline-specific skill bindings with priority
* fix: resolve cherry-pick conflicts for agentskills onto sandbox
- Remove non-existent external_kb service import
- Add skill_mgr mock to localagent sandbox_exec tests
- Keep database version at 24 (sandbox branch's latest)
* feat(skills): upgrade to package-backed skills with sandbox execution
Evolve the skills system from pure prompt-based to package-backed with
sandbox tool execution support:
- Add source_type/package_root/entry_file/skill_tools fields to Skill entity
- SkillManager loads SKILL.md from local package directories
- SkillToolLoader as 4th dispatch layer in ToolManager (query-scoped)
- LocalAgent injects skill tools into use_funcs on skill activation
- BoxService.execute_skill_tool() runs scripts in sandbox (ro mount, env params)
- Skill tool names auto-namespaced as skill__{skill}__{tool}
- API validation for package_root allowlist and entry path traversal
- Frontend source_type toggle, package_root input, skill_tools editor
- Migration renumbered to 025 with ALTER TABLE fallback for existing DBs
- Fix unclosed limitation section in i18n files
- Fix skills API methods misplaced outside BackendClient class
* fix: test info
* feat(skills): switch skills to package-backed storage and add import tooling
- skills 从 inline/package 双轨收敛成 package-first
- instructions 改为写入并读取 SKILL.md
- 新增本地目录扫描和 GitHub 安装 skill
- 前端把 skills 整合进 plugins 页，新增 SkillsComponent 和 GitHub 导入弹窗
- skill form 去掉 source_type / type 筛选，改成目录扫描驱动
- Box skill tool 挂载模式从 ro 改成 rw
- 测试和中英文文案同步更新
* feat: simplify langbot skill create and import
* refactor(skills): clean up legacy skill API and harden activation flow
* refactor(skills): remove skill dependency expansion and add skill_get
* fix: lint
* fix: delete
* fix(skills): align tool manager loader initialization
* refactor: remove sandbox execute skill
* fix(skills): hide activation markers and isolate skill activation flow
* refactor(skills): switch skill model to filesystem-backed packages
* refactor(skills): switch skill model to filesystem-backed packages
* refactor(skills): unify runtime skill access around filesystem paths
* refactor(skills): unify runtime skill access around filesystem paths
* feat(skills): align rw package design and fix skill activation, visibility, and lint issues
* refactor(skills): replace rich authoring API with import/reload flow and update
Box design doc
* feat(box): add sandbox_exec tool loop for local-agent calculations
* feat(box): add host workspace mounting and sandbox_exec guidance
* feat(box): add BoxProfile with resource limits and improved output truncation
- Implement head+tail output truncation (60/40 split) so LLM sees both
beginning and final results; add streaming byte-limited reads in backend
to prevent unbounded memory usage (_MAX_RAW_OUTPUT_BYTES = 1MB)
- Define BoxProfile model with locked fields and max_timeout_sec clamping
- Add four built-in profiles: default, offline_readonly, network_basic,
network_extended with differentiated resource and security constraints
- Add resource limit fields to BoxSpec (cpus, memory_mb, pids_limit,
read_only_rootfs) and pass corresponding container CLI flags
(--cpus, --memory, --pids-limit, --read-only, --tmpfs)
- Profile loaded from config (box.profile), applied in service layer
before BoxSpec validation; locked fields cannot be overridden by
tool-call parameters
* feat(box): add obs
* refactor(box): unify box service lifecycle and local runtime
management
* refactor(box): remove legacy in-process runtime code and clean up smells
After the architecture settled on always using an independent Box Runtime
service, several pieces of compatibility code and design shortcuts were
left behind. This commit cleans them up:
- Remove `LocalBoxRuntimeClient` and `create_box_runtime_client` from
production code (moved to test-only helper).
- Remove unused `_clip_bytes` method from backend.
- Remove `__langbot_session_placeholder__` hack by making `BoxSpec.cmd`
default to empty and validating non-empty only in `runtime.execute()`.
- Extract `get_box_config()` helper to eliminate 5× duplicated config
access boilerplate.
- Remove `session_id`/`host_path`/`host_path_mode` from the LLM-facing
tool schema to enforce request-scoped session isolation.
- Fix dual shutdown path: `NativeToolLoader.shutdown()` no longer calls
`box_service.shutdown()` (handled by `Application.dispose()`).
- Simplify `_assert_session_compatible` with a loop.
- Inline client creation in `BoxRuntimeConnector`.
- Remove redundant `BOX__RUNTIME_URL` env var from docker-compose
(auto-detected by code).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(box/mcp): integrate MCP stdio with Box sandbox — auto-isolation, dep install, security
## Summary
When Podman/Docker is available, all stdio-mode MCP servers now automatically
run inside Box containers with dependency installation, path rewriting, and
lifecycle management. When no container runtime exists, LangBot starts normally
and stdio MCP falls back to host-direct execution.
## What changed
### MCP stdio → Box integration (mcp.py)
- Add `MCPServerBoxConfig` pydantic model for structured box configuration
with validation and defaults (network, host_path_mode, timeouts, resources)
- Auto-infer `host_path` from command/args with venv detection: recognizes
`.venv/bin/python` patterns and walks up to the project root
- Rewrite host paths to container `/workspace` paths transparently
- Replace venv python commands with container-native `python`
- Auto-detect `pyproject.toml`/`setup.py`/`requirements.txt` and run
`pip install` inside the container before starting the MCP server
- Copy project to `/tmp` before install to handle read-only mounts
- Add retry with exponential backoff (3 retries, 2s/4s/8s delays)
- Add Box managed process health monitoring (poll every 5s)
- Fix session leak: `_cleanup_box_stdio_session()` now runs in `finally`
block of `_lifecycle_loop`, covering all exit paths
- Fix retry logic: `_ready_event` is only set after all retries exhaust
or on success, not on first failure
- Enhance `get_runtime_info_dict()` with `box_session_id` and `box_enabled`
### Box security (security.py — new)
- `validate_sandbox_security()` blocks dangerous host paths:
`/etc`, `/proc`, `/sys`, `/dev`, `/root`, `/boot`, `/run`,
docker.sock, podman socket
- Called at the start of `CLISandboxBackend.start_session()`
### Box models (models.py)
- Add `BoxHostMountMode.NONE` — skips volume mount entirely
- Adjust `validate_host_mount_consistency` to allow arbitrary workdir
when `host_path_mode=NONE`
### Box backend (backend.py)
- Add `validate_sandbox_security()` call in `start_session()`
- Add `langbot.box.config_hash` label on containers for drift detection
- Handle `BoxHostMountMode.NONE` — skip `-v` mount arg
- Add `cleanup_orphaned_containers()` to base class (no-op default) and
CLI implementation (single batched `rm -f` command)
### Box runtime (runtime.py)
- Call `cleanup_orphaned_containers()` during `initialize()` to remove
lingering containers from previous runs
### Box service (service.py)
- Graceful degradation: `initialize()` catches runtime errors and sets
`available=False` instead of crashing LangBot startup
- Add `available` property and guard on `execute_sandbox_tool()`
- Add `skip_host_mount_validation` parameter to `build_spec()` and
`create_session()` — MCP paths are admin-configured and trusted,
bypassing `allowed_host_mount_roots` restrictions meant for
LLM-generated sandbox_exec commands
### Default behavior
- stdio MCP servers automatically use Box when `box_service.available`
is True (Podman/Docker detected); no explicit `box` config needed
- When no container runtime exists, falls back to host-direct stdio
- MCP Box defaults: `network=on` (for pip install), `read_only_rootfs=false`
(for site-packages), `host_path_mode=ro`, `startup_timeout=120s`
### Tests
- `test_box_security.py`: blocked paths, safe paths, subpath rejection
- `test_mcp_box_integration.py`: config model, path rewriting, venv
unwrap, host_path inference, payload building, runtime info, box
availability check
- `test_box_service.py`: `BoxHostMountMode.NONE` validation tests
* feat(box/mcp): instance-based orphan cleanup, error classification, session API, and integration tests
## Changes
### Precise orphan container cleanup
- Runtime generates a unique instance_id on startup
- Every container gets a `langbot.box.instance_id` label
- `cleanup_orphaned_containers()` only removes containers from
previous instances, preserving containers owned by the current one
- Containers from older versions (no label) are also cleaned up
- `cleanup_orphaned_containers` added to `BaseSandboxBackend` as
a no-op default method, removing hasattr duck-typing
### Fine-grained MCP error classification
- New `MCPSessionErrorPhase` enum with 7 phases: session_create,
dep_install, process_start, relay_connect, mcp_init, runtime,
tool_call
- Each phase in `_init_box_stdio_server()` sets the error phase
before re-raising, enabling precise failure diagnosis
- `retry_count` tracked across retry attempts
- `get_runtime_info_dict()` exposes `error_phase` and `retry_count`
### GET /v1/sessions/{id} API
- `BoxRuntime.get_session()` returns session details including
managed process info when present
- `handle_get_session` HTTP handler + route in server.py
- `BoxRuntimeClient.get_session()` abstract method + remote impl
### stdio defaults to Box when runtime is available
- `_uses_box_stdio()` checks `box_service.available` instead of
requiring explicit `box` key in server_config
- `BoxService.initialize()` catches runtime errors gracefully,
sets `available=False` instead of crashing LangBot startup
- When no container runtime exists, stdio MCP falls back to
host-direct execution
### Code quality (from /simplify review)
- Extracted `_VENV_DIRS` / `_VENV_BIN_DIRS` module-level constants
- Removed dead `_box_network_mode()` method and unused `bc` variable
- Fixed broken import `from ....box.models` → `from ...box.models`
- Cached `_resolve_host_path()` result — computed once, passed through
- Config hash now includes `host_path` field
- Batched orphan cleanup into single `rm -f` command
### Session leak fix
- `_cleanup_box_stdio_session()` now runs in `_lifecycle_loop`'s
finally block, covering all exit paths (normal shutdown, error,
retry, final failure)
### Integration tests
- 6 end-to-end tests covering managed process lifecycle, WebSocket
stdio bidirectional IO, session cleanup verification, single
session query, process exit detection, and orphan cleanup safety
* refactor: use rpc
* fix: import
* refactor(box): clean up sandbox subsystem code quality and efficiency
- Fix O(n²) stderr trimming in runtime.py with running length tracker
- Remove dead code: RESERVED_CONTAINER_PATHS, _subprocess_wait_task,
unused config_hash computation, unused imports
- Deduplicate connection callback in BoxRuntimeConnector, parse URL once
- Use enum comparison instead of stringly-typed spec.network.value check
- Replace manual _result_to_dict/_session_to_dict with model_dump()
- Cache NativeToolLoader tool definition and sandbox system guidance
- Extract _is_path_under() helper to eliminate duplicated path checks
- Import SANDBOX_EXEC_TOOL_NAME from native.py instead of redefining
- Add JSON startswith guard in logging_utils to skip futile json.loads
- Fix ruff lint errors (F401 unused imports, F841 unused variables)
* fix: ruff
* refactor(sandbox): keep box logic out of pipeline and localagent
- Move sandbox system-prompt guidance from LocalAgentRunner into
BoxService.get_system_guidance() so all box domain knowledge stays
in the box module.
- Remove standalone logging_utils.py; merge format_result_log() into
MessageHandler base class alongside cut_str().
- Strip sandbox-specific JSON parsing from log formatting; tool
results now use generic truncation.
- Revert TYPE_CHECKING changes in stage.py and runner.py that were
unrelated to this feature.
- Skip two test files affected by a pre-existing circular import
(runner ↔ app) until the import cycle is resolved in a separate PR.
* refactor(box): move box runtime to langbot-plugin-sdk
Extract self-contained box runtime modules (actions, backend, client,
errors, models, runtime, security, server) to langbot-plugin-sdk and
update all imports to use `langbot_plugin.box.*`. Keep only service
and
connector in LangBot core as they depend on the Application context.
- Update docker-compose to use `langbot_plugin.box.server` entry
point
- Update pyproject.toml to use local SDK via `tool.uv.sources`
- Remove migrated source files and their unit/integration tests
- Update remaining test imports to match new module paths
* fix: ruff
* fix(box): tighten sandbox exposure and restore box integration coverage
* refactor(types): remove quoted annotations under postponed evaluation
* chore(sandbox): move MCP loader changes to follow-up branch
* refactor(plugins): simplify GitHub install flow to default master archive
* revert(api): restore plugin GitHub import flow in plugins controller
* Improve data-root handling and skill install previews
* Add managed skill authoring tools for local agents
* Refactor the skills UI around sidebar detail pages
* Document why managed skill authoring tools bypass box
* fix: lint
* feat(web): refactor plugin/skill install flows and fix skills page
- Fix sidebar skill icon
- Add skills route and error page component
- Refactor plugin GitHub install from dialog modal to inline card
- Add skill install dropdown menu (create/upload/github) in sidebar
- Wire sidebar → skills page communication via pendingSkillInstallAction context
- Add i18n keys for error page and skill install actions
* fix(web): persist sidebar collapsible section open state on navigation
Sections opened via sub-item navigation now retain their expanded state
when the user switches to a different section, instead of collapsing
because the isActive fallback becomes false.
---------
Co-authored-by: youhuanghe <1051233107@qq.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Junyan Qin <rockchinq@gmail.com>
* feat(sandbox): add MCP box integration on top of sandbox base (#2083)
* refactor(mcp): extract box stdio runtime helper
* refactor(box): introduce reusable workspace session helper
* refactor(box): run Box Runtime as subprocess inside LangBot container
Remove the separate langbot_box_runtime Docker service. Box Runtime
now always launches as a local stdio subprocess, regardless of whether
LangBot runs in Docker or not. The WebSocket transport path is kept
only for explicit runtime_url configuration (remote deployment).
This simplifies deployment by eliminating cross-container path mapping
and network hops. Box Runtime is a pure scheduling process (talks to
Docker socket / nsjail), it does not execute user code or touch the
filesystem, so container isolation is unnecessary — unlike Plugin
Runtime.
* fix(web): prevent first-emission snapshot from swallowing unsaved changes in pipeline editor
When switching runner (e.g. local-agent → n8n), the newly mounted stage's
first emit would re-capture the saved snapshot, erasing the dirty state
caused by the runner change. The save button would incorrectly go dim.
- Skip snapshot re-capture in handleDynamicFormEmit when form is already dirty
- Add mount-time emit to N8nAuthFormComponent (matching DynamicFormComponent)
- Use stable onSubmitRef to prevent useEffect subscription churn
- Add previousInitialValues guard to prevent initialValues echo loops
* style(web): align plugin list header button heights
* docs(review): update Box architecture review documents
Replace old review docs with 5 focused documents:
- box-architecture.md: deep architecture analysis (LangBot + SDK)
- box-issues.md: 22 issues rated P0/P1/P2
- box-test-coverage.md: test coverage analysis
- box-tob-analysis.md: toB commercialization analysis
- box-vs-plugin-runtime.md: Box vs Plugin runtime comparison
* feat(web): improve login error layout and add Terms of Service link
- Improve backend connection error display with bordered container,
inline icon, and better visual hierarchy
- Extract actual error message from axios response object
- Add Terms of Service link (https://langbot.app/terms) to login footer
- Add termsOfService i18n key for all 7 locales
* refactor(web): replace all hardcoded SVG icons with lucide-react
Unify icon usage across the entire frontend by replacing 67 hardcoded
SVG icons with lucide-react components across ~25 files. This improves
consistency, maintainability, and reduces bundle duplication.
Key replacements:
- Sidebar nav: Zap, LayoutDashboard, Bot, Workflow, BookMarked, etc.
- MCP forms: Loader2, XCircle, Trash2
- Monitoring: Sparkles, MessageSquare, CheckCircle2, RefreshCw, etc.
- Cards: Clock, Star, Workflow, Hexagon, Puzzle, Github, etc.
- Misc: Paperclip, AudioLines, CloudUpload, Layers, Heart, Smile
Zero hardcoded <svg> tags remain in .tsx files.
* fix(web): stop polling plugin tasks when no active installs
The PluginInstallTaskProvider was unconditionally polling
getAsyncTasks every 3s on all /home/* routes. Now it only
syncs once on mount and starts periodic polling only when
there are active (non-terminal) install tasks.
* fix(deps): update langbot-plugin version and add new dependencies
* refactor: use Space API for release checks and stop idle polling
- version.py: switch release list API from GitHub to space.langbot.app,
remove unused in-place update logic (update_all, compare_version_str),
translate all comments/logs to English
- PluginInstallTaskContext: only poll when active install tasks exist
* feat(box): add --standalone-box flag and 3-way transport decision for Box runtime
Align Box runtime connection logic with Plugin runtime's pattern:
- Docker: WebSocket to langbot_box container (ws://langbot_box:5411)
- --standalone-box: WebSocket to external Box process (ws://localhost:5411)
- Windows: subprocess + WebSocket (workaround for async stdio limitation)
- Unix/macOS: subprocess + stdio pipe (unchanged)
BoxRuntimeConnector now inherits ManagedRuntimeConnector for subprocess
lifecycle reuse. Add langbot_box service to docker-compose.yaml.
* refactor(box): use single port with path-based routing for Box WS
Update connector to use ws://host:5410/rpc/ws instead of ws://host:5411.
Update review docs to reflect the single-port architecture.
* feat(web): show Box runtime status in plugin debug info popover
Add Box status section to the debug info popover on the plugin list page,
displaying connection status, backend info, profile, active sessions,
and recent error count. Fetched from GET /api/v1/box/status in parallel
with plugin debug info. Includes i18n for all 8 supported languages.
* fix(web): remove ephemeral sandbox count from Box status display
The active_sessions count reflects transient sandbox containers that
expire after 5 minutes of inactivity, making it misleading in the UI.
Keep only connection status, backend, profile, and error count.
* feat(box): configurable sandbox scope and unified skill containers
Replace the per-message session_id with a template-based system
configurable per pipeline via 'Sandbox Scope' in the local-agent panel.
Default scope is per-chat ({launcher_type}_{launcher_id}).
Unify skill exec into the same container as default exec — skills are
mounted at /workspace/.skills/{name}/ via extra_mounts instead of
getting separate containers. All pipeline-bound skills are injected
at container creation time.
- Add box-session-id-template to pipeline metadata (select, 4 options, 8 languages)
- Add resolve_box_session_id() and build_skill_extra_mounts() to BoxService
- Rewrite native.py skill exec path to use execute_tool with shared session
- Update tests for new session_id format
- Add design doc: docs/review/box-session-scope.md
* feat(web): show active sandbox details in Box status popover
Display sandbox count and a detailed list of active sessions including
session ID, image, backend, resources (CPU/memory), network mode, and
last used time. Fetched from GET /api/v1/box/sessions in parallel.
Includes i18n for all 8 supported languages.
* feat(box): add startup and availability logging for sandbox tools
Log Box runtime initialization result (success with profile info, or
failure warning). Log native tool availability status at ToolManager
startup so it's immediately clear whether exec/read/write/edit tools
are registered for the LLM.
* feat(box): support custom sandbox container image via config.yaml
Add 'image' field to box config section. When set, it overrides the
profile default image (python:3.11-slim) for all sandbox containers.
Priority: caller-specified > config.yaml image > profile default.
* feat(box): add heartbeat and reconnection for Box runtime connector
Add 20-second heartbeat ping loop to detect silent Box runtime
disconnections. On disconnect, set available=false and attempt
reconnection after 3 seconds via the disconnect callback chain.
- BoxRuntimeConnector: heartbeat loop, disconnect callback parameter,
disconnect detection in connection callback and WS failure handler
- BoxService: wire disconnect callback to toggle available state and
re-initialize the connector on reconnection
* feat(web): move runtime status to dashboard, clean up plugin debug popover
Add SystemStatusCards component to the monitoring dashboard showing
Plugin Runtime and Box Runtime connection status with details (backend,
profile, sandbox count). Remove all Box/session status from the plugin
page debug popover — it now only shows debug URL and key.
Includes i18n for all 8 supported languages.
* refactor(web): compact system status into a single card alongside metrics
Replace the separate two-card row with a single compact 'System Status'
card placed as the 5th column in the metrics grid. Shows green/red dots
for Plugin Runtime and Box Runtime. Click to expand a popover with
connection details (backend, profile, sandbox count).
* feat: show connector error details for Plugin and Box runtime status
Record Box connector error in BoxService and expose it as
'connector_error' in GET /api/v1/box/status when unavailable.
Display error messages in the dashboard System Status popover
for both Plugin Runtime (plugin_connector_error) and Box Runtime
(connector_error) when they are disconnected.
* fix(web): auto-refresh system status and show disconnect errors in real time
Poll Plugin Runtime and Box Runtime status every 30 seconds so the
dashboard reflects disconnections without a manual page refresh.
Also re-fetch when the popover is opened for immediate feedback.
* fix(box): handle RPC failure in get_status/get_sessions gracefully
When the Box runtime disconnects, there is a race between the heartbeat
flipping _available=false and the frontend polling get_status(). If the
poll arrives first, client.get_status() throws a ConnectionClosedError
which propagated as a 500, causing the frontend to show a grey dot
(null status) instead of a red dot with error details.
Now get_status() catches RPC errors and returns available=false with
the exception message as connector_error. get_sessions() returns an
empty list when unavailable or on RPC failure.
* fix(box): add persistent reconnection loop with exponential backoff
The previous disconnect handler only retried once and then gave up.
Now spawns a background task that retries with exponential backoff
(3s, 6s, 12s, ... up to 60s) until the Box runtime is reachable again.
Uses a _reconnecting guard to prevent duplicate loops. Calls
connector.dispose() before each retry to clean up stale tasks.
* fix(box): detect disconnect when handler.run() returns normally
The generic Handler.run() catches ConnectionClosedError and breaks out
of its loop (normal return) instead of raising, because it has no
disconnect_callback. The old code only triggered reconnection in the
except branch, so a clean WebSocket close was never detected.
Now treat handler.run() returning normally (after successful handshake)
as a disconnect event, triggering the reconnection callback.
* fix(web): refresh system status card when clicking Refresh Data button
Pass a refreshKey prop through OverviewCards to SystemStatusCard that
increments on each Refresh Data click, triggering a re-fetch of Plugin
and Box runtime status alongside the monitoring data refresh.
* fix(web): fix system status card stuck in loading state
fetchStatus(showLoading=false) never called setLoading(false), so the
initial loading=true was never cleared. Simplify to always setLoading
in the finally block — the spinner only shows on the very first load
since subsequent fetches complete near-instantly.
* feat(web): show active sandbox details in dashboard Box status popover
Fetch box sessions alongside status and display each active sandbox
in the popover with session ID, image, resources (CPU/memory), and
last used time.
* feat(box): add global sandbox scope option
Add a 'Global (shared by all)' option to the sandbox scope selector.
Uses a constant '{global}' template variable that always resolves to
'global', so all users and chats share one sandbox container.
* refactor(web): replace popover with dialog for system status details
Replace the dropdown popover with a proper Dialog for runtime status
details. Add a small info button on the System Status card that opens
the dialog. Session details now show in a spacious 2-column grid layout
with full image name, backend, CPU/memory, network, mount path, and
created/last-used timestamps.
* fix(web): widen system status dialog and fix scroll border issue
Use max-w-2xl (matching other dialogs) instead of max-w-lg. Move
overflow-y-auto to an inner container with overflow-hidden on
DialogContent to prevent padding bleed at scroll edges.
* feat(web): add tooltips for truncated fields in system status dialog
Wrap session_id, image, and mount path fields with Tooltip components
so hovering over truncated text shows the full value.
* feat: add download button
* feat: successfully install
* feat: delete old filter
* feat: youhua frontend
* fix: align box runtime launch args
* feat: translate
* feat: refactor market
* feat: youhua qianduan
* chore: rename extension zh translation
* feat(extensions): unify extensions endpoint and refresh extensions page UX
- Rename /home/plugins route to /home/extensions and update all sidebar links.
- Add unified GET /api/v1/extensions returning plugins, MCP servers and skills,
sorted by name; replace the three separate frontend fetches with this single call.
- Migrate the extensions page to shadcn primitives (Tabs/Card/Alert/Badge/Skeleton/
Switch/Label) and clean up hardcoded color tokens on the extension card.
- Add a localStorage-persisted "Group by type" switch that, when enabled in the
All Types tab, renders extensions grouped by type with a compact section header.
- Show a spinner while loading and rename the empty-state copy from
"No plugins installed" to "No extensions installed".
- Rename the "格式 / Formats" filter label to "类型 / Types" across all 8 locales.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(extensions): fallback lucide icon when extension icon is missing
Render a tinted lucide icon (Puzzle / Server / Sparkles) on the extension
card when the icon URL is empty or the image fails to load. Picked icons
distinct from EventListener (AudioWaveform) and KnowledgeEngine (Book) to
avoid visual collision with plugin component badges.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(sidebar): unify installed-extensions list with plugins, MCP and skills
- Render plugins, MCP servers and skills together under the "Installed
Extensions" sidebar entry, alphabetically sorted to match the list page.
- Resolve per-item routes by extension type (plugin -> /home/extensions,
mcp -> /home/mcp, skill -> /home/skills) and gate the plugin-only hover
context menu on extensionType === 'plugin'.
- Lift the "group by type" toggle into SidebarDataContext (still persisted
in localStorage) so the sidebar groups items with section headers
whenever the list page has the toggle enabled.
- Show lucide fallback icons (Server / Sparkles / Puzzle) tinted in the
LangBot blue for MCP, skill, and missing-icon plugin items, overriding
the SidebarMenuSubButton svg color rule.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(extensions): mobile-friendly layout for extensions and add-extension pages
- Stack the extensions page header vertically on small screens, let the
filter Tabs scroll horizontally if they overflow, hide the debug
button label below sm and let the install/debug controls wrap.
- Constrain the debug popover and its inputs to the viewport width so
they no longer overflow on phone-sized screens.
- Drop the card grid from a fixed 30rem column to a min(100%, 22rem)
column at base / 28rem at sm, and reduce the gap, so cards render
cleanly at 360px+ widths in both flat and grouped views.
- Make the add-extension header actions wrap on lg- viewports and the
install dialog responsive instead of a hard 500px box.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat: change ui
* feat: delete version for mcp and skills
* fix: constrain home page content width
* fix: preserve monitoring card borders under sticky filters
* fix(box): restore sandbox config and shared mcp runtime
* fix(box): harden sandbox session isolation
* fix(skill): remove auto activation setting
* feat(skill): align skill system with Claude Code's Tool Call design
- Replace text marker activation with `activate` tool (Tool Call mechanism)
- Replace 7 authoring tools with 2: `activate` + `register_skill`
- Add builtin skills loading from templates/skills/
- Add create-skill as first builtin skill
- Remove SKILL_ACTIVATION_MARKER and text detection methods
- Tool Result returns SKILL.md content (protects KV Cache)
This aligns with Claude Code's progressive disclosure pattern:
- Metadata (name+description) always visible in tool description
- SKILL.md body loaded on activate via Tool Call
- Bundled resources accessible through virtual path mapping
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* feat(tools): add glob and grep native sandbox tools
Add file discovery and content search capabilities to the sandbox:
- glob: Find files by pattern (supports ** recursive matching)
- grep: Search file contents with regex patterns
Both tools respect skill package paths and include safety limits
(max 100 files for glob, max 200 matches for grep).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* feat(skill): add skill file browsing capability
- Add API endpoints for listing/reading/writing skill files
- Add FileTree component in SkillForm for directory browsing
- Users can now view scripts/, references/, assets/ directories
- Files can be selected and edited in the instructions textarea
- Add translations for new file browsing features
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(skill): copy builtin skills to data/skills on startup
- Builtin skills (templates/skills/) are now copied to data/skills/
- Users can view and manage builtin skills in the UI
- Rename SkillAuthoringToolLoader to SkillToolLoader
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(skill): improve file browsing and fix path handling
- Fix nested directory display in skill file tree (preserve root entries)
- Fix file content display when clicking files in skill browser
- Add skill manager and tool manager as proper package modules
- Separate fileContent state to allow editing non-SKILL.md files
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(toolmgr): correct skill_tool_loader attribute name
Rename skill_authoring_tool_loader to skill_tool_loader in execute_func_call
and shutdown methods to match the attribute defined in initialize().
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(native): update tool descriptions to use register_skill
Replace references to removed import_skill_from_directory with
register_skill in exec/write/edit tool descriptions.
* feat(toolmgr): enhance tool initialization with backend availability checks
* refactor: remove unused imports and clean up code in various files
* feat: polish extension detail pages
* feat: persist sidebar list expansion
* fix: refine extension ui and backend errors
* fix: align add extension marketplace ui
* feat: manage skills through box runtime
* feat: support github skill installation
* fix: import github skill directories
* feat: install market extensions from card click
* feat(web): improve skill import flow
* feat: polish extension import flow
* fix(mcp): stabilize shared box managed processes
* fix(web): improve backend retry and sidebar scrolling
* docs(review): refresh box architecture review for feat/sandbox
Sync the docs/review/ suite to the current state of the feat/sandbox branch
(both LangBot and langbot-plugin-sdk), ~30 commits ahead of the prior review.
- box-architecture.md: rewrite for the new box.{backend,runtime,local,e2b}
config schema, add E2B backend, 6 native tools (incl. glob/grep), Skill
Tool Call activation, shared multi-process MCP container, SkillManager,
BoxSkillStore (SDK), 25 actions, 9 error types, heartbeat/reconnect
- box-issues.md: move resolved items (reconnect, heartbeat, Windows, nsjail
image conflict, frontend monitoring card) into a Resolved section; add
new P0 (INIT/backend ordering), P1 (extra_mounts immutability after
container creation), P2 (skill_store test gap, integration tests not in CI)
- box-session-scope.md: add §0 Implementation Status — Phase 1 shipped,
MCP unification landed earlier than originally scoped
- box-test-coverage.md: realign file inventory (4,400 -> 6,500 LOC),
add 7 new test files including SDK backend_selection/e2b/skill_store
- box-tob-analysis.md: connection recovery now满足基本要求; add E2B and
backend self-heal to capabilities; tick off Phase 1 reconnect/heartbeat
- box-vs-plugin-runtime.md: heartbeat/reconnect/Windows support now aligned
with Plugin Runtime; revise remaining gaps (WS auth, shared base class)
* refactor(box): use unified env-override mechanism for box.local config
The box module hand-rolled its own LANGBOT_BOX_LOCAL_* env parsing in two
places (connector._get_box_config and service._local_config), duplicating
logic that LoadConfigStage._apply_env_overrides_to_config already provides
generically via the SECTION__SUBSECTION__KEY convention.
- Drop the bespoke LANGBOT_BOX_LOCAL_* parsing; read box.local straight
from instance_config (the unified BOX__LOCAL__* overrides are already
applied before BoxService initializes)
- Harden _load_allowed_mount_roots to accept a comma-separated string,
since the generic mechanism stores a freshly-created key as a raw
string when config.yaml has no box.local.allowed_mount_roots entry
- docker-compose: rename the langbot container env vars to
BOX__LOCAL__* (the canonical convention); remove them entirely from
the langbot_box container — the Box runtime never reads box.local from
env/config.yaml, it is configured via the INIT RPC action
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* test: repair stale skill/sandbox tests for feat/sandbox
The skill subsystem moved to Tool-Call activation and a Box-managed
skill store; several tests still asserted removed APIs and a sys.modules
stub leaked across the suite. Full unit suite now green (was 23 failing).
- test_skill_tools: drop TestSkillManagerActivation (text-marker API
removed); rewrite TestSkillActivationHelper around the current
skill.activation.register_activated_skill; replace the CRUD
TestSkillAuthoringToolLoader with TestSkillToolLoader covering the
current activate/register_skill tools and sandbox-availability gating
- test_tool_manager_native: ToolManager attr is skill_tool_loader (not
skill_authoring_tool_loader); native loader now exposes 6 tools
(exec/read/write/edit/glob/grep) and requires initialize() with a
backend-available get_status()
- test_localagent_sandbox_exec: remove obsolete activation-marker
leakage tests and their helper providers
- test_model_service / pipeline conftest: give the mocks skill_mgr=None
so PreProcessor's local-agent skill-binding guard short-circuits
- test_n8nsvapi: stop permanently overwriting sys.modules
('langbot.pkg.provider.runner' etc.); save and restore around the
import so other modules get the real LocalAgentRunner base class
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* ci(tests): run unit tests on every push to feat/** branches
- Add feat/** to push branches so long-lived feature branches are
tested on every push (they accumulate large changes before a PR)
- Drop the push path filter entirely: every push to master/develop/
feat/** now runs the full unit suite (the old 'pkg/**' filter never
matched the real source path 'src/langbot/pkg/**', so backend-only
pushes silently skipped tests)
- Fix the same broken path glob on the pull_request trigger
('pkg/**' -> 'src/langbot/pkg/**')
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(skill): harden mount/reload paths and HTTP errors against stale skill cache
The Box backends behave inconsistently when extra_mounts reference a
missing host directory (nsjail aborts the entire sandbox start, Docker
silently creates a root-owned empty dir on the host, E2B silently skips
the upload). The cache in skill_mgr.skills is only refreshed on
in-process mutations, so out-of-band changes — container rebuilds,
manual rm in the box volume, anything the LangBot API didn't drive —
leave a stale skill that later produces one of those bad mount paths.
- box/service.py: build_skill_extra_mounts now filters skills whose
package_root is not isdir on the LangBot-visible filesystem and logs
a warning, instead of passing the bad mount through to the backend
- skill/manager.py: reload_skills (Box path) drops skills whose
package_root is missing on the LangBot-side filesystem before they
reach the in-memory cache, with a summary warning
- api/http/controller/groups/skills.py: file/CRUD handlers now also
catch BoxError (RuntimeError subclass, previously slipping past
``except ValueError`` and surfacing as 500); list/get handlers gain
a try/except so a transient Box RPC failure becomes a clean 400
instead of a stack trace
Tests added for build_skill_extra_mounts (skip missing, skip empty,
no skill manager) and SkillManager.reload_skills (drop missing on Box
path). Full unit suite: 279 passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(box): add box.enabled toggle and gate consumers on availability
Make the Box sandbox runtime optional. When ``box.enabled`` is false in
config (or when an enabled Box fails to connect), every dependent feature
degrades to the same disabled-state UX rather than crashing or silently
falling back to less safe code paths.
Backend:
- config.yaml: new top-level ``box.enabled: true`` flag (default true)
- BoxService:
- Read box.enabled on construction
- initialize() short-circuits when disabled — no remote WS connect, no
stdio subprocess fork
- _on_runtime_disconnect is a no-op when disabled (no reconnect loop
on a deliberately-off service)
- get_status() now exposes ``enabled`` so the frontend can tell
"disabled in config" from "configured but failed"
- MCP stdio loader (mcp_stdio.uses_box_stdio): requires box_service to
be available, not just installed
- MCP _init_stdio_python_server: when ap.box_service exists but is
unavailable, refuse the stdio server with an actionable error instead
of silently falling through to host-stdio (which bypasses the sandbox
the operator asked for). Setups without ap.box_service installed at
all keep the legacy host-stdio fallback for pre-Box dev mode
- SkillService._require_box_for_write: refuses create/update/install/
write_skill_file when ap.box_service is installed but unavailable.
Distinguishes disabled vs failed in the error message so the UI can
surface the right hint. Legacy setups (no ap.box_service) keep the
local fallback path — that distinction is what keeps the existing
local-skills tests valid
Tests:
- Box disabled-state behavior (4 cases)
- Skill write refusal in disabled & failed states (7 cases)
- MCP stdio runtime info policy updated to match new refuse-when-down
behavior
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(web): surface Box disabled/unavailable state across consumers
When Box is disabled in config (``box.enabled = false``) or fails to
connect, every dependent UI surface now degrades visibly:
- ``useBoxStatus`` hook: shared, polled 30s, exposes ``available``,
``disabled`` (config-off) and a single ``hint`` key so callers don't
have to re-derive the three states
- ``BoxUnavailableNotice`` reusable Alert banner driven by that hint
- Dashboard SystemStatusCards: three-state dot + label
(connected / disabled-gray / disconnected-red); disabled state shows
the ``boxDisabled`` hint, failed state continues to show the connector
error. Plugin block kept untouched
- Skills page (create view) and SkillDetailContent (edit view):
Save button disabled and banner inserted above the form when Box is
unavailable — matches the backend gate added in the previous commit
- PipelineExtension skill section: ``enable_all_skills`` switch, Add
Skill button and Remove buttons all gate on Box availability;
banner inline under the section header
- PipelineFormComponent: banner above the ``local-agent`` stage card
when Box is unavailable, since that stage carries the sandbox-bound
``box-session-id-template`` field
- Box status payload type (``ApiRespBoxStatus.enabled``) and 8 locale
files updated with ``boxDisabled`` / ``boxUnavailable`` /
``boxRequiredHint`` strings
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs(box): document the box.enabled toggle and gate behavior matrix
- docker-compose: move ``langbot_box`` under compose profiles
(``box`` and ``all``) so ``docker compose up`` no longer requires
the sandbox container. Inline comment explains how to pair the
profile choice with ``box.enabled`` so the langbot service does not
thrash trying to reach a runtime that was never started
- docs/review/box-architecture.md:
- Annotate ``box.enabled`` in the config.yaml example, listing the
exact side effects (no remote/stdio connect; tools/skills/MCP
stdio off; reads still work)
- Replace the bare compose snippet with the actual profile-driven
invocation and the BOX__ENABLED pairing
- New "关闭/连接失败时的行为矩阵" section: a single table mapping
every consumer (native tools, activate/register_skill, stdio MCP,
skill list/CRUD, pipeline AI config, extensions page, dashboard)
to its disabled-state behavior, plus the legacy ``ap.box_service``
distinguisher note
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* refactor(pipeline-form): swap Box banner for field-level disable_if + tooltip
The previous commit hard-coded a BoxUnavailableNotice banner above the
``local-agent`` stage card. That works, but it shouts at the user about
every field in that stage when in reality only one field —
``box-session-id-template`` — depends on the sandbox.
Use the dynamic-form schema's existing variable-injection mechanism
(``__system.*`` references via ``systemContext``) and add a sibling to
``show_if``: ``disable_if`` + ``disabled_tooltip``. The field stays
visible, becomes inert, and an info icon next to its label exposes the
reason on hover. The rest of the AI tab is left untouched.
- entities/form/dynamic.ts: extend IDynamicFormItemSchema with
``disable_if: IShowIfCondition`` and ``disabled_tooltip: I18nObject``
- DynamicFormComponent: evaluate disable_if with the same resolver as
show_if; OR the result into isFieldDisabled; render an Info tooltip
trigger next to the label when the condition matches
- ai.yaml metadata: attach disable_if (__system.box_available eq false)
and a localized disabled_tooltip to box-session-id-template
- PipelineFormComponent: drop the BoxUnavailableNotice import and the
per-stage banner; pass ``systemContext={ box_available: boxAvailable }``
only for the local-agent stage so other stages aren't paying the
re-render cost
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(mcp): friendly UI message when stdio MCP refused by Box state
Previously the MCP detail dialog dumped the raw RuntimeError text from
``_init_stdio_python_server`` — English-only, prefixed with "Failed
after 4 attempts", and exposing internal config names. The retry
wrapper also kept retrying a refusal that is deterministically going
to fail again, polluting logs.
Replace the raw text with a structured signal:
- New ``MCPSessionErrorPhase.BOX_UNAVAILABLE`` enum value. The stdio
refusal path sets it before raising and uses a short opaque
discriminator (``box_disabled_in_config`` / ``box_unavailable``) as
the message body — never user-facing
- ``_lifecycle_loop_with_retry`` short-circuits on
``BOX_UNAVAILABLE``: surfaces the error immediately, no retries,
no "Failed after N attempts" prefix. Silences the warning storm
seen during smoke-testing
- ``MCPServerRuntimeInfo`` (TS type) now declares ``error_phase``,
``retry_count``, ``box_session_id``, ``box_enabled`` to match what
the backend already returns in get_runtime_info_dict()
- Both MCP detail forms (``mcp/components/mcp-form/MCPForm.tsx`` and
``plugins/mcp-server/mcp-form/MCPFormDialog.tsx``) detect
``error_phase === 'box_unavailable'`` and render a two-line
localized notice: state line ("Box disabled / unreachable") plus
remediation line ("enable Box or switch to http/sse")
- 8 locale files (en/zh-Hans/zh-Hant/ja/ru/vi/th/es) get
``mcp.boxDisabledStdioRefused``, ``mcp.boxUnavailableStdioRefused``,
``mcp.boxStdioRefusedSuggestion``
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(mcp-web): block stdio MCP creation at the form when Box is unavailable
When Box is disabled in config (``box.enabled = false``) or unreachable,
saving a new MCP server in stdio mode produced one that could never
start — the user would only learn that from the runtime error on the
detail page. Stop the user before they save instead.
Both MCP forms (the page-level ``MCPForm.tsx`` and the older dialog
``MCPFormDialog.tsx``) now:
- Disable the ``stdio`` option in the mode select when Box is
unavailable, with a small "(requires Box)" suffix so the reason is
obvious. Existing stdio configs still display their current value
- Show ``BoxUnavailableNotice`` inline under the mode select when the
currently-selected mode is stdio and Box is unavailable, so editing
a stale stdio config makes the cause visible
- Disable the Save / Submit button while stdio is selected under that
condition. ``MCPForm`` exposes a new ``onSaveBlockedChange`` prop
so the parent ``MCPDetailContent`` can disable both its Submit and
Save buttons. ``MCPFormDialog`` disables its Save button locally
- Refuse the submit handler too (Enter-key path) with a toast carrying
the same i18n message
i18n: ``mcp.boxRequired`` (short tag in the disabled option) and
``mcp.stdioBlockedByBoxToast`` added to all 8 locales.
Backend runtime gate (``_init_stdio_python_server`` refusal +
``BOX_UNAVAILABLE`` error_phase + retry short-circuit) stays in place
as the last line of defence for API bypass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(web): prevent plugin config form overflow
* refactor(skill): remove all local-filesystem fallbacks; Box is the sole source
Skills now flow exclusively through the Box runtime. Every read and write
method funnels through ``_box_service()``; when Box is unavailable
(disabled in config, connection failed, or simply not installed) the
operation either returns an empty surface (``list_skills`` → []) or
raises with a clear ``Box runtime ... not initialised / disabled /
unavailable: ...`` message via the new ``_require_box(action)`` helper.
Why: the legacy local-fallback path scanned ``data/skills/``, but Box
manages its own ``box.local.skills_root`` (default ``data/box/skills/``).
The two diverging directories caused stale / phantom skill lists when
Box flapped, and the local-fallback writes silently bypassed all the
sandboxing the operator had configured.
SkillService (``api/http/service/skill.py``):
- New ``_require_box(action)`` returns the box service or raises a
structured ValueError. ``_require_box_for_write`` kept as alias
- ``list_skills`` → returns [] when Box is down so the UI can render
the disabled banner cleanly
- ``get_skill`` / ``get_skill_by_name`` → return None
- All read-file / write-file / scan-dir / create / update / delete /
install / preview methods → ``_require_box`` then box delegate.
Local fallback bodies (shutil.copytree, tempfile.mkdtemp, preview
pipelines) removed entirely
SkillManager (``pkg/skill/manager.py``):
- ``reload_skills`` returns early with empty cache when Box is down.
data/skills/ discovery loop removed
- ``refresh_skill_from_disk`` now just reports cache presence; the
on-disk re-parse is gone since Box is the only writer
Tests:
- Drop 11 obsolete test_skill_service.py tests that exercised the
removed local-fallback paths (create/install/file/delete/update)
- Add list-empty + read-refused tests; flip the legacy-allow test to
legacy-refuses-too
- Rewrite refresh_skill_from_disk test to match the new behaviour
Several helper methods (_managed_skill_path, _resolve_skill_path,
_preview_skill_candidates, _install_preview_candidates, etc.) are now
unreachable; a follow-up commit will prune them so this diff stays
reviewable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* chore(skill): prune dead local-filesystem helpers left over from Box migration
Follow-up to the Box-only refactor. The previous commit removed the
local-fallback BRANCHES from every public method; this one removes the
HELPERS those branches called, which are now unreachable.
SkillService (service/skill.py): 787 → 449 lines
Removed: scan_directory (sync), _read_skill_package, _write_skill_md,
_resolve_create_field, _managed_skill_path,
_managed_install_root_for_package, _normalize_package_root,
_resolve_skill_path, _find_skill_entry, _discover_skill_directories,
_safe_extract_zip, _extract_uploaded_skill_to_temp,
_download_github_skill_to_temp, _resolve_github_source_root,
_build_preview_target_dir, _preview_skill_candidates,
_select_preview_candidates, _install_preview_candidates,
_preview_source_root, _resolve_installed_skills, plus the
module-level _FRONTMATTER_FIELDS and _build_skill_md.
Kept (still needed by the surviving GitHub-import path):
_download_github_asset, _download_github_skill_directory_as_zip,
_find_github_skill_archive_entry, _copy_github_skill_directory_to_zip,
_is_github_skill_md_url, _parse_github_skill_md_url,
_resolve_github_skill_md_package_name, _validate_github_asset_url,
_uploaded_skill_target_stem, _validate_skill_name.
Imports dropped: shutil, tempfile, yaml, ....utils.paths.
SkillManager (skill/manager.py): 187 → 88 lines
Removed: get_managed_skills_root, _discover_skill_directories,
_find_skill_entry, _load_skill_file, _normalize_package_root.
Imports dropped: datetime, parse_frontmatter, paths.
Tests:
- test_skill_service.py: drop the 3 sync scan_directory tests +
skill_service fixture + _create_skill_file helper
- test_skill_tools.py: drop test_load_skill_file_success; rename
TestSkillManagerPackageLoading → TestSkillManagerCache
Full unit suite: 277 passed, 1 skipped. ``ruff check`` clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(skill): re-inject skill index into local-agent system prompt
The contributor's original PR (#1917) appended an ``Available Skills``
index to the system prompt before the LLM saw the user message, so the
LLM could decide whether to activate a skill. ``7145447b`` removed the
text-marker activation flow and, together with it, the entire system
prompt injection — but the Tool Call replacement only put the available
skills inside the ``activate`` tool's description. In practice the LLM
ignores tool descriptions for selection and goes straight to native
tools, so user-visible skill activation silently broke.
Restore the injection, adapted for the Tool Call era:
- SkillManager regains ``get_skill_index(bound_skills)`` and
``build_skill_aware_prompt_addition(bound_skills)``. The addendum
carries only ``name (display_name): description`` for each
pipeline-visible skill plus one instruction line pointing at the
``activate`` tool. No SKILL.md contents — KV cache stays clean
- PreProcessor appends the addendum to the first system message (or
inserts a new one) of ``query.prompt.messages`` for the local-agent
runner. Handles plain-string and ContentElement[] bodies. Skips
cleanly when no skills are visible
- 3 new test_preproc cases: injection happens, bound-skills subset
honoured, empty addendum touches nothing. 280 passed
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(box): downgrade get_status.available when backend probed unavailable
Until now ``BoxService.get_status`` returned ``available: true`` whenever
the runtime connector was healthy, even if the runtime itself reported
``backend: { available: false }`` (operator selected nsjail without the
binary, Docker daemon crashed mid-session, E2B credentials wrong, ...).
The dashboard / ``useBoxStatus`` hook / skill_service gate consumed the
top-level flag and showed "connected" while every actual call to native
exec or skill management would fail.
The native-tool loader already polled ``status.backend.available``
independently and hid its tools correctly, but every other consumer
(dashboard banner, the disabled-state hint, the LLM-facing message)
disagreed with it.
Combine the two in the payload: ``available = self._available AND
status.backend.available``. When ``backend.available`` is false we now
also surface a ``connector_error`` that names the backend
("Configured sandbox backend \"nsjail\" is unavailable") so the dialog
shows the actionable reason instead of an empty error pane. The
detailed ``backend`` object is preserved unchanged for the dialog.
Internal ``box_service.available`` (used by ``skill_service`` writes,
``mcp_stdio.uses_box_stdio``, the reconnect callback) is intentionally
NOT changed — it still tracks connector health only, so a backend blip
does not trigger spurious reconnect loops.
Tests:
- ``test_get_status_downgrades_available_when_backend_dead`` — exercise
the new branch (connector OK, backend.available=false → top-level
available=false, connector_error mentions the backend name)
- ``test_get_status_keeps_available_true_when_backend_ok`` — guard
against regressing the happy path
Live-verified with ``box.backend: nsjail`` on macOS (no nsjail binary):
``GET /api/v1/box/status`` now returns ``available: false`` with the
named connector_error, instead of the previous misleading
``available: true``.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(web): surface the specific Box failure reason in unavailable banner
When Box is configured but the runtime reports its backend is dead
(e.g. ``box.backend = nsjail`` but the binary is missing, or Docker
daemon crashed), the backend now returns a structured
``connector_error`` like ``Configured sandbox backend "nsjail" is
unavailable``. The previous notice only said "Box sandbox is
unavailable" + a generic "enable Box" hint, hiding the actionable
detail.
- ``useBoxStatus``: derive ``reason`` from ``status.connector_error``.
Only exposed for the failed-state (``hint === 'boxUnavailable'``),
since the disabled-by-config message already carries its reason
- ``BoxUnavailableNotice``: insert the reason as a small monospaced
line between the state message and the action hint. The disabled
variant is unchanged (operator chose the state)
- Wire ``reason`` through every existing call site (Skills page +
detail, PipelineExtension, both MCP forms). Old unused ``context``
prop dropped
Net layout (3 lines, still compact):
⚠ Box sandbox is unavailable — sandbox tools, skill add/edit, ...
Configured sandbox backend "nsjail" is unavailable
This feature requires the Box runtime. Enable it in config ...
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* test: reconcile master's unit tests with feat/sandbox refactors
The merge from master brought in new unit tests that target pre-refactor
APIs on feat/sandbox. Reconcile each:
- factories/app.py: FakeApp now exposes a Mock skill_mgr (with empty .skills
dict + inert prompt-addition builder) and a Mock pipeline_service so the
PreProcessor skill-index injection branch can run end-to-end in tests.
- pipeline/conftest.py: eagerly import langbot.pkg.pipeline.pipelinemgr so
pipeline.stage is fully initialised before any individual stage test
(preproc, longtext, ...) tries to lazy-load it. Without this preload,
running test_preproc.py in isolation hit a circular-import error via the
stage -> app -> pipelinemgr -> stage chain.
- provider/test_tool_manager.py: ToolManager now probes four loaders
(native -> plugin -> mcp -> skill). Inject inert native + skill mocks in
the execute_func_call fixture and assert all four shutdowns fire.
- utils/test_paths.py: drop the three cwd-dependent _check_if_source_install
cases. The refactor walks Path(__file__).resolve().parents looking for
pyproject.toml + main.py, so cwd no longer factors in and there's no
file read to mock-fail. The positive case and caching test still apply.
- utils/test_version.py: delete entirely. is_newer and compare_version_str
were removed when VersionManager was refactored to use the Space API for
release checks (1b4107a9); the tests targeted a surface that no longer
exists.
* refactor(box): launch box runtime via the lbp CLI subcommand
Mirror the plugin runtime: box is now started through the same CLI entry
point (langbot_plugin.cli) instead of the box module directly.
- docker-compose.yaml: langbot_box command runs `langbot_plugin.cli ... box`
(WebSocket is the default transport, no flag needed — matches `rt`).
- box/connector.py: both subprocess launch sites (_start_local_stdio and
the Windows _start_subprocess_then_ws path) invoke
`langbot_plugin.cli.__init__ box`, using `-s` for the stdio transport.
- docs/review: update stale `-m langbot_plugin.box[.server]` references.
Pairs with the SDK change that removes box's direct-launch entry points
(python -m langbot_plugin.box / .box.server) and the legacy --mode flag.
* chore: bump langbot-plugin beta 1
* fix(ci): resolve langbot-plugin from PyPI and clear lint failures
CI on feat/sandbox failed across Unit Tests, Lint and Build Dev Image.
Root causes and fixes:
- pyproject.toml had a [tool.uv.sources] editable override pinning
langbot-plugin to ../langbot-plugin-sdk. That path only exists in a
paired local checkout, so `uv sync` failed on every CI runner
("Distribution not found"). Remove the override and regenerate uv.lock
so langbot-plugin==0.4.0b1 resolves from PyPI, matching master.
- tests/integration/api/test_pipelines.py: the pipeline extensions
endpoint now calls ap.skill_service.list_skills(); add the missing
skill_service mock to the fake_pipeline_app fixture (the test came
from master, the endpoint change from feat/sandbox).
- Apply ruff format to three src files and prettier to three web files
that had committed formatting drift, failing `ruff format --check`
and `pnpm lint`.
* chore: bump beta version
* docs: remove BOX_BACKEND override reference
* fix(pipelines): stop attributing dashboard debug WS to bound web_page_bot
The dashboard pipeline-debug WebSocket
(/api/v1/pipelines/<uuid>/ws/connect) and the embed widget WebSocket
(/api/v1/embed/<bot_uuid>/ws/connect) already live on separate paths,
but the debug handler ran `_find_owner_bot(pipeline_uuid)` and, when
the same pipeline happened to be bound to a web_page_bot, passed that
bot as `owner_bot` into `handle_websocket_message`. The adapter then
used the page bot's listeners + adapter for the request, so debug
sessions were logged as "page bot" activity in the dashboard.
Debug sessions must always run under the built-in websocket_proxy_bot.
Remove `_find_owner_bot`, drop the `owner_bot` parameter from the
debug-path `_handle_receive`, and call `handle_websocket_message`
without it so the adapter takes its default proxy-bot branch. The
embed handler still resolves and passes its `runtime_bot` for the
page-bot path, so attribution there is unchanged.
* fix(plugin): install marketplace MCP from canonical mode + extra_args
_install_mcp_from_marketplace read the dropped `mcp_data.config` field
and reconstructed mode/extra_args by guessing from the URL — which lost
stdio's command/args/env/box entirely, so stdio MCP installs from the
marketplace always failed.
Use the Space record's canonical `mode` and `extra_args` directly (the
same shape stored in mcp_servers), and gate the install on `mode`
instead of the removed `config`. After a successful install, best-effort
POST to the marketplace install endpoint to bump install_count.
* feat(web): show recommendation lists in plugin market; mixed-type icons
The marketplace recommendation lists (curated rows from Space) were never
mounted in the plugin market page. Wire them in:
- fetch recommendation lists on mount and render them above the extension
grid, only when no search/filter is active.
Recommendation lists now mix plugins, MCPs and skills, so resolve each
card's icon by type (plugin / mcp / skill marketplace icon URL) instead of
always using the plugin icon endpoint.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(web): auto-open install dialog from one-click deep link
Accept a deep link from LangBot Space's one-click install:
/home/add-extension?install=1&extension_type=<plugin|mcp|skill>&author=&name=&version=
On mount, populate the install info, open the confirm dialog directly, and
strip the params from the URL. Reuses the existing marketplace install flow.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat: push marketplace URL to runtime; fix market client base race
- On connecting to the plugin runtime, push the configured space.url via the
new SET_RUNTIME_CONFIG action so the runtime downloads plugins from the same
Space, instead of relying on its own CLOUD_SERVICE_URL env/default. Wrapped
in try/except so an older SDK without the action degrades gracefully.
- web: the plugin market fetched recommendation lists (and listings) via the
sync cloud client before its baseURL was resolved from system info, so it
hit the default space.langbot.app. Await getCloudServiceClient() before the
initial fetches and for the recommendation list.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(web): don't show MCP "connection failed" while still connecting
The MCP status UI rendered "连接失败" for any non-connected state, so during a
normal connection attempt the subtitle showed "连接失败" while the status pill
below it showed "连接中..." — contradictory.
Only treat an explicit ERROR (or box-unavailable) status as failed; a
CONNECTING or initial/unresolved status now shows "连接中". Applied to the MCP
detail form (subtitle + StatusDisplay) and the MCP server card.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(web): type-aware install dialog + refresh sidebar after install
The marketplace install confirm dialog was hardcoded to "安装插件 / 确定要安装
插件 X 吗" for every type. Make it type-aware (plugin / MCP / skill) and show
more info: type chip, author/name id, and version when present.
Also refresh all sidebar extension lists (plugins, MCP servers, skills) when
an install task completes, so the newly-installed extension appears
immediately regardless of type (previously only refreshPlugins ran).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(web): richer install dialog (icon + name + description), drop redundant type row
The install dialog already states the type in its title, so the "类型" row was
redundant. Replace the info box with the extension's icon (avatar), display
name, author/name id + version, and description — built from the PluginV4 for
in-app installs and from the icon endpoint by type for the one-click deep link.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(web): TDZ crash in add-extension (installIconURL before installInfo)
installIconURL was computed above the useState declaration of installInfo,
causing "Cannot access 'installInfo' before initialization" (500) on the
add-extension page. Move the computation below the state declarations.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(web): redesign install-progress dialog for MCP/skill
The progress dialog showed plugin-only stages (download + dependency install)
for every type. MCP/skill have no such steps, so show a single
"installing → done/failed" row for them (MCP: adding & connecting the server;
skill: installing the package) while keeping the detailed download/deps
stages for plugins.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(web): add missing market.componentName i18n keys
The marketplace component filter (and component badges) used
market.componentName.{Tool,Command,EventListener,KnowledgeEngine,Parser,Page}
but those keys only existed under plugins.componentName, so the market UI
showed raw keys. Add a componentName block to the market namespace (zh-Hans +
en-US; other locales fall back to zh-Hans).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(web): sidebar extensions refresh button + full-name tooltip
- Add a refresh button to the installed-extensions category header in the
sidebar; it re-fetches plugins + MCP servers + skills and spins while
loading.
- The sidebar item tooltip now shows the extension's full name (with the
description below when present), so truncated MCP/extension names are
readable on hover.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(plugin-market): rename component filter to "插件组件" with hint tooltip + persist filters
- Rename the in-app plugin market component filter label to "插件组件" /
"Plugin Component"
- Add an Info icon tooltip explaining what plugin components are (Tool /
Command / EventListener, etc.)
- Persist filter selections (type / component / tags / sort) in localStorage
so they survive reloads; restored on mount (URL type param still wins)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(plugin-market): restore missing "页面"(Page) component filter option
The market component-filter list on this branch was a diverged rewrite that
dropped the Page component kind master had added. The i18n key
(market.componentName.Page) already existed; re-add the Page entry to the
componentOptions list so plugins providing Page components can be filtered.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(i18n): reword plugin component filter hint
Drop the redundant "插件组件是" lead-in and mention that components extend
LangBot's capabilities; mirror the wording in en-US.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(i18n): backfill missing market/addExtension keys in 6 locales
check-i18n surfaced that market.componentName.*, market.filterByComponentHint
and the addExtension.install* keys existed only in en-US/zh-Hans. Backfill
them for es-ES, ja-JP, ru-RU, th-TH, vi-VN and zh-Hant (reusing each locale's
existing component-name translations) and align the filterByComponent label
with the new "Plugin Component" wording. check-i18n now passes for all locales.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* i18n(plugins): relabel "group by type" as "group by format"
The installed-extensions grouping is by extension format (plugin / MCP / skill),
so rename the toggle label accordingly across all 8 locales (key unchanged).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(plugin-market): cursor-pointer on tag filter trigger
The TagsFilter Select trigger used the default cursor; add cursor-pointer so the
tag filter is clearly clickable.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(sidebar): show edition badge (Community / Cloud) in logo area
Add a small badge next to the LangBot name in the sidebar header that reflects
systemInfo.edition: a neutral "Community" badge for the community edition and a
blue "Cloud" badge for the cloud edition. Adds sidebar.editionCommunity /
sidebar.editionCloud across all 8 locales.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* i18n(sidebar): unify zh-Hans cloud edition label to 云端版
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(sidebar): edition badge - drop hover, use "Cloud" in all locales
The edition badge is not interactive, so remove the hover background on the
cloud badge. Also use the literal "Cloud" label uniformly across all locales
instead of localized variants (云端版/クラウド版/...).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(box): cap tool-call loop and run workspace-quota walk off the event loop
Two robustness fixes that bite under normal sandbox usage (not just attack),
hardening the self-hosted community edition before release:
- localagent: cap the tool-call loop at MAX_TOOL_CALL_ROUNDS (128). A looping
or adversarial model could otherwise emit tool calls indefinitely (each
potentially a sandbox exec), producing a non-terminating request and runaway
cost. The cap is generous enough not to interrupt legitimate multi-step
agentic workflows.
- box.service: make _enforce_workspace_quota async and run the recursive
workspace scan via asyncio.to_thread. It ran on every quota-enforced exec and
a large workspace would block the whole asyncio runtime (all bots/pipelines).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(review): refresh box docs; trim issue list to SaaS blockers only
Community self-hosted edition is release-ready, so the box review docs are
updated to current state (date 2026-06-02 + status note) and box-issues.md is
rewritten to keep only the SaaS / multi-tenant / network-exposed release
blockers (S1-S8): unauthenticated control plane, no per-pipeline exec
authorization, unbounded sessions + no reaper, no kernel-level quota, mount
validation gaps (/ + extra_mounts), missing container hardening, lock-around-
cold-start, and the lower-severity follow-ups. Resolved items (tool-call loop
cap, async quota scan, host_path mount allowlist, _is_path_under dedup) moved to
a short "resolved before community release" record; community-only and
pure-cleanup items dropped.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* chore(deps): pin langbot-plugin to 0.4.0
Track the stable SDK release (0.4.0b1 -> 0.4.0); regenerate uv.lock.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: WangCham <651122857@qq.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: fdc310 <82008029+fdc310@users.noreply.github.com>
Co-authored-by: Junyan Qin <rockchinq@gmail.com>
32 KiB
32 KiB
Box 系统架构深度分析
更新日期: 2026-06-02 状态更新: 自部署社区版已具备发布条件（box 可选、降级完善、无迁移欠债）；工具调用循环上限、配额遍历异步化、
host_path挂载白名单等已落地。剩余多租户 / 安全硬化项见 SaaS 阻塞项清单。 分支:feat/sandbox(LangBot + langbot-plugin-sdk) 相关文档: SaaS 阻塞项 | Session 作用域 | Runtime 对比 | 测试覆盖 | toB 分析
1. 全局架构
┌──────────────────────────────────────────────────────────────────┐
│ LangBot 主进程 │
│ │
│ LocalAgentRunner ──> ToolManager ──> NativeToolLoader │
│ │ │ │ │
│ │ │ exec / read / write / edit │
│ │ │ glob / grep │
│ │ │ │
│ │ ├──> MCPLoader ──> BoxStdioSession │
│ │ │ (shared 容器, 多 process) │
│ │ │ │
│ │ ├──> SkillToolLoader (activate 工具) │
│ │ │ │
│ │ ├──> SkillAuthoringToolLoader │
│ │ │ │
│ │ └──> PluginToolLoader │
│ │ │
│ BoxService (门面) │
│ ├─ Profile 管理 (locked 字段) │
│ ├─ Host mount 校验 (allowed_mount_roots) │
│ ├─ Workspace quota 检查 │
│ ├─ 输出截断 (head+tail) │
│ ├─ Session ID 模板解析 (resolve_box_session_id) │
│ ├─ 技能挂载组装 (build_skill_extra_mounts) │
│ ├─ 重连循环 (_reconnect_loop, 指数退避) │
│ └─ BoxRuntimeConnector │
│ ├─ 心跳 loop (20s ping) │
│ └─ ActionRPCBoxClient │
│ │ Action RPC (stdio 或 WebSocket) │
│ │
│ SkillManager (skill_mgr) │
│ └─ 从 Box runtime 拉取 skills, 不可用时回落 data/skills │
└──────────────────────────────────────────────────────────────────┘
│
▼
┌──────────────────────────────────────────────────────────────────┐
│ Box Runtime 进程 (SDK 侧) │
│ │
│ BoxServerHandler (Action RPC 处理, INIT 配置注入) │
│ │ │
│ BoxRuntime (session 管理 / 进程生命周期 / TTL reaper) │
│ │ └─ session.managed_processes: dict[pid, _ManagedProcess]
│ │ │
│ Backend (启动时根据 box.backend 配置选择): │
│ DockerBackend ──┐ │
│ PodmanBackend ──┤── CLISandboxBackend │
│ NsjailBackend ──┘ (本地 CLI 或 fallback 到容器内 CLI) │
│ E2BBackend (云沙箱, 需要 E2B_API_KEY) │
│ │
│ BoxSkillStore │
│ ├─ list / get / create / update / delete │
│ ├─ scan_skill_directory / read_skill_file / write_skill_file │
│ └─ preview_skill_zip / install_skill_zip (zip 或 GitHub) │
│ │
│ aiohttp 单端口服务 (默认 :5410): │
│ /rpc/ws — Action RPC │
│ /v1/sessions/{id}/managed-process/ws — 默认 process │
│ /v1/sessions/{id}/managed-process/{pid}/ws — 指定 process │
└──────────────────────────────────────────────────────────────────┘
│
▼
┌──────────────────────────────────────────────────────────────────┐
│ 容器 / 沙箱 (Docker/Podman 容器, nsjail sandbox, 或 E2B 远程沙箱) │
│ - 隔离文件系统 / 网络 / PID 命名空间 │
│ - 资源限制 (CPU, 内存, PID 数, 可选 workspace 配额) │
│ - 主挂载 (host_path → mount_path) + 任意条 extra_mounts │
│ └─ Skills 通过 extra_mounts 挂在 /workspace/.skills/<name> │
│ - exec: 用户命令在此执行 │
│ - managed process: 多个长驻进程并存 (MCP Server / 自定义服务) │
└──────────────────────────────────────────────────────────────────┘
核心设计原则:
- Box Runtime 作为独立进程运行，通过 Action RPC 与 LangBot 主进程通信，两者复用 SDK 的 IO 层（Handler → Connection → Controller）
- 一个 session_id 对应一个容器/沙箱实例。同一 session 内可并存多条 mount 与多个 managed process
- Skill / 默认 exec / MCP Server 共享同一个 session 容器（详见 box-session-scope.md）
2. LangBot 侧模块
2.1 BoxService (pkg/box/service.py, 722 行)
应用层门面，协调 Profile、安全校验、配额、连接、Skill 挂载与 Session 模板：
主要公开方法（按定义顺序）：
BoxService
├─ initialize() 连接 Box Runtime + 默认 workspace 准备
├─ _on_runtime_disconnect(connector) 触发重连
├─ _reconnect_loop(connector) 指数退避重连
├─ available (property) 连接状态
│
├─ resolve_box_session_id(query) 从 pipeline 模板解析 session_id
├─ build_skill_extra_mounts(query) 组装 pipeline-bound skill 的挂载列表
│
├─ execute_tool(parameters, query) Agent 调用 exec 时的入口
│ ├─ _apply_profile / build_spec
│ ├─ _validate_host_mount
│ ├─ _enforce_workspace_quota (phase=pre)
│ ├─ client.execute(spec)
│ ├─ _enforce_workspace_quota (phase=post)
│ └─ _truncate (stdout/stderr)
│
├─ execute_spec_payload(spec_payload, ...) 内部入口（其他 loader 调用）
├─ create_session(spec_payload, ...) 显式创建 session
├─ start_managed_process(session_id, ...) 启动 managed process
├─ get_managed_process(session_id, pid) 查询进程状态（pid 默认 'default'）
├─ stop_managed_process(session_id, pid) 单独停止某个 managed process
├─ get_managed_process_websocket_url(...) 返回 WS attach URL
│
├─ list_skills() / get_skill(name) Skill 元数据
├─ create_skill / update_skill / delete_skill Skill CRUD
├─ scan_skill_directory(path) 扫描目录
├─ list_skill_files / read_skill_file / write_skill_file
├─ preview_skill_zip / install_skill_zip zip / GitHub 安装
│
├─ shutdown() / dispose() 清理：RPC SHUTDOWN + 进程终止
├─ get_status() / get_sessions() / get_recent_errors()
└─ get_system_guidance() LLM 系统提示
Profile 系统: 4 个内置 Profile（default / offline_readonly / network_basic / network_extended），locked frozenset 字段不可被 LLM 覆盖。参数合并顺序：Profile defaults → LLM 请求参数 → locked 强制值。
输出截断: 默认 4000 字符上限，保留前 60% + 后 40%，中间插入 [...truncated...]。
Skill 挂载合并: execute_tool() 调用时，build_skill_extra_mounts(query) 会把当前 pipeline-bound 的所有 skill 的 package_root 作为 extra_mounts 加入 BoxSpec，挂在 /workspace/.skills/<name>。LLM 通过 activate 工具显式激活某个 skill 后，工具调用才允许引用这个 skill 的虚拟路径。
2.2 BoxRuntimeConnector (pkg/box/connector.py, 357 行)
管理与 Box Runtime 的通信连接：
- 本地 stdio: Unix/macOS 默认路径，fork
python -m langbot_plugin.cli.__init__ box -s --ws-control-port {port}子进程（与 plugin runtime 统一走lbpCLI 入口） - 本地 subprocess + WS: Windows 本地（asyncio ProactorEventLoop 不支持 stdio pipe）
- 远程 WebSocket: Docker 部署 /
box.runtime.endpoint显式配置时，连接ws://{host}:{port}/rpc/ws - 同步等待:
asyncio.Event+wait_for(timeout=30s)模式确认连接 - 心跳:
_heartbeat_loop()每 20s 调用ping()，失败仅 DEBUG 日志（断开检测靠 connection close） - 重连:
runtime_disconnect_callback由 BoxService 提供，触发_reconnect_loop - INIT 注入: 连接建立后立即下发当前
box.*配置子树（剔除runtime私有字段），Runtime 据此初始化 backend
历史改进: 2026-04-16 版本本文档曾列 P0 「Box 无心跳 / 无重连」，已修复（commit
2dfd9d5d、c6882cf、5029d9c等）。
2.3 BoxWorkspaceSession 工具 (pkg/box/workspace.py, 413 行)
此文件目前提供两类能力：
- 路径与命令重写工具函数 —
normalize_host_path/rewrite_mounted_path/unwrap_venv_path/rewrite_venv_command/infer_workspace_host_path，被 MCP loader 与 Skill 路径解析共用。 BoxWorkspaceSession— 围绕 BoxService 的轻量包装，专供 MCP-in-Box 场景使用（管理一个共享 session 的 session_id、构建挂载 payload、stage host 文件到共享 workspace）。
变化点: 早期 Skill exec 会为每个 skill 创建独立 BoxWorkspaceSession（独占 session）；当前实现已转为 extra_mounts 模式，Skill 不再独占容器，只追加挂载。这部分 wrapping 逻辑已从 native loader 移除。
2.4 policy.py (pkg/box/policy.py, 98 行) — 仍是死代码
三层安全策略设计（SandboxPolicy / ToolPolicy / ElevatedPolicy），全项目无任何导入或调用。详见 SaaS 阻塞项 S2。
2.5 SkillManager (pkg/skill/manager.py, 186 行)
SkillManager
├─ initialize() 调用 reload_skills()
├─ reload_skills() 先从 Box runtime list_skills()，
│ 不可用则回落 data/skills/ 扫描
├─ refresh_skill_from_disk() 单 skill 重新加载
├─ get_skill_by_name(name)
└─ get_managed_skills_root() 返回 Box 视角的 skills_root 路径
skill 元数据通过 parse_frontmatter 解析 SKILL.md 头部（name / description / instructions），不再做整体扫描的代价（典型 < 50 个）。
2.6 Skill activation (pkg/skill/activation.py, 33 行) + Skill loader 辅助
历史上 skill 通过 LLM 在文本中输出 [ACTIVATE_SKILL:name] 标记激活；当前已改为 Tool Call 机制：
SkillToolLoader(pkg/provider/tools/loaders/skill.py, 157 行) 暴露activate工具，参数为 skill 名- 工具实现调用
register_activated_skill(query, skill_data)，将激活态写入query.variables['_activated_skills'] - 这种 KV-cache-friendly 模式对齐 Claude Code 设计；详见 box-session-scope.md §4.3 的 Tool Call 描述
activation.py 现仅保留对外辅助函数（pipeline 层调用 loader 的 register_activated_skill）。
3. SDK 侧模块
3.1 BoxRuntime (box/runtime.py, 599 行)
核心编排器，管理 session 生命周期与 backend 调度：
Session 生命周期:
Client EXEC / CREATE_SESSION
│
▼
_get_or_create_session(spec)
├─ _reap_expired_sessions_locked() 清理 TTL 过期 session
├─ 已存在? → _assert_session_compatible() → 复用
├─ Backend session 失踪? → 重建 (commit c6882cf)
└─ 新建? → backend.start_session(spec) → 创建容器
│ └─ 应用 spec.extra_mounts （多挂载）
▼
execute(spec)
├─ 获取 session lock (每 session 独立)
├─ backend.exec(session, spec) 在容器中执行命令
├─ 更新 last_used_at
└─ 超时? → 销毁 session
│
▼
Session 保持存活直到:
├─ TTL 过期 (默认 300s，下次操作时清理)
├─ 执行超时 (自动销毁)
├─ 客户端 DELETE_SESSION
└─ SHUTDOWN
关键设计:
- 每 session 有独立
asyncio.Lock，同一 session 内的命令串行执行 - 每 session 维护
managed_processes: dict[process_id, _ManagedProcess]，支持多个长驻进程并存（MCP / 自定义） - 全局
_lock保护_sessionsdict 的读写 - 兼容性检查：比较核心 spec 字段，
image字段对不支持自定义镜像的 backend（nsjail/E2B）会跳过
Backend 选择 (_select_backend): 优先级
- 显式
box.backend配置（docker/nsjail/e2b） local(默认) → Docker / Podman / nsjail CLI 顺序探测get_status调用时若当前 backend 不可用，会尝试重新选择 (commite5617c7)
3.2 Backend 系统
CLISandboxBackend (box/backend.py, 411 行)
Docker / Podman 公共基类：
start_session(spec):
1. validate_sandbox_security(spec)
2. docker/podman run -d --rm --name <name>
--network none (可选)
--cpus/--memory/--pids-limit
--read-only + --tmpfs /tmp
-v <host>:<mount>:<mode> 主挂载
-v <extra.host>:<extra.mount>:.. 额外挂载 (extra_mounts)
<image> sh -lc 'while true; do sleep 3600; done'
3. 返回 BoxSessionInfo
exec(session, spec):
docker/podman exec -e KEY=VAL <container>
sh -lc 'mkdir -p <workdir> && cd <workdir> && <cmd>'
start_managed_process(session, spec):
docker/podman exec -i <container>
sh -lc 'mkdir -p <cwd> && cd <cwd> && exec <command> <args>'
返回 asyncio.subprocess.Process (stdin/stdout PIPE)
容器以 idle 进程启动，实际命令通过 docker exec 执行。--rm 确保容器退出时自动清理。
Windows 支持: backend 内对 Windows 路径处理与 subprocess 调用做了适配（commit 120817a）。
孤儿清理: 启动时枚举 langbot.box=true 标签的容器，instance_id 不匹配的强制删除。
NsjailBackend (box/nsjail_backend.py, 552 行)
轻量级 Linux 沙箱（无容器引擎依赖）：
- 使用 namespace 隔离（user/mount/pid/ipc/uts/cgroup/net）
- 挂载宿主
/usr//lib//bin//sbin只读 + 选定/etc条目 - 每 session 创建独立目录（workspace/tmp/home）
- 资源限制: cgroup v2 优先，fallback 到 rlimit
- CLI 兼容: 通过
shutil.which(self._nsjail_bin)检测系统安装版 nsjail；不存在时再尝试容器内 nsjail（commit686fcc0、feed530） - 无自定义镜像: 使用宿主 OS，
image字段固定为'host'，兼容性检查跳过 image
E2BBackend (box/e2b_backend.py, 429 行)
云沙箱后端（commit 75b547f 引入）：
- 通过
e2bSDK 与 E2B 平台通信 - 配置：
box.e2b.api_key/api_url/template - 支持
extra_mounts（commit0fea9b1同步上传文件） - 无本地容器引擎依赖，适合无 Docker 的部署或 SaaS 多租户场景
- 不支持自定义 image 字段，由 template 控制
3.3 Server (box/server.py, 508 行)
单端口 aiohttp 服务（默认 5410），通过路径区分（commit 8c71ec5 合并端口）：
- Action RPC (
/rpc/ws):BoxServerHandler处理所有 action，包括INIT配置注入、skill store 操作等 - WS Relay (
/v1/sessions/{id}/managed-process/ws与/v1/sessions/{id}/managed-process/{pid}/ws): 双向桥接 WebSocket ↔ 指定 managed process stdin/stdout
stdio 模式同样会在 5410 启动 aiohttp，专门承担 managed process attach；Action RPC 走 stdin/stdout。
3.4 Client (box/client.py, 377 行)
ActionRPCBoxClient 封装 Handler.call_action() 调用：
- 25+ 方法对应 25+ 个 RPC action（exec / session / managed-process / skill / status / shutdown）
- 错误还原:
_translate_action_error()通过字符串前缀匹配还原 SDK 侧异常类型 execute()timeout = 300s，其他默认 15sBoxRuntimeClient是 ABC，供后续可能的非 RPC 实现复用
包级别 __init__.py 显式导出：BoxRuntimeClient、ActionRPCBoxClient（commit df9c722）。
3.5 Actions (box/actions.py, 34 行)
LangBotToBoxAction 枚举共定义 25 个 action：
| 类别 | Actions |
|---|---|
| 控制 | INIT、HEALTH、STATUS、GET_BACKEND_INFO、SHUTDOWN |
| 执行 | EXEC |
| Session | CREATE_SESSION / GET_SESSION / GET_SESSIONS / DELETE_SESSION |
| Managed Process | START_MANAGED_PROCESS / GET_MANAGED_PROCESS / STOP_MANAGED_PROCESS |
| Skill | LIST_SKILLS / GET_SKILL / CREATE_SKILL / UPDATE_SKILL / DELETE_SKILL / SCAN_SKILL_DIRECTORY / LIST_SKILL_FILES / READ_SKILL_FILE / WRITE_SKILL_FILE / PREVIEW_SKILL_ZIP / INSTALL_SKILL_ZIP |
3.6 Models (box/models.py, 331 行)
核心数据模型：
| 模型 | 用途 |
|---|---|
BoxNetworkMode |
OFF / ON |
BoxExecutionStatus |
COMPLETED / TIMED_OUT |
BoxHostMountMode |
NONE / READ_ONLY / READ_WRITE |
BoxManagedProcessStatus |
RUNNING / EXITED |
BoxMountSpec |
单条挂载（host_path/mount_path/mode）— 新增 |
BoxSpec |
执行请求；新增 extra_mounts: list[BoxMountSpec]、persistent、workspace_quota_mb |
BoxProfile |
4 个内置 Profile + locked frozenset |
BoxSessionInfo |
Session 状态（含 backend_name/created_at/last_used_at） |
BoxManagedProcessSpec |
长驻进程参数（process_id/command/args/env/cwd） |
BoxManagedProcessInfo |
进程状态（status/exit_code/stderr_preview/attached） |
BoxExecutionResult |
执行结果（status/exit_code/stdout/stderr/duration_ms） |
BoxSpec 校验器: workdir 默认继承 mount_path；host_path 支持 POSIX 和 Windows 路径；设置 host_path 时 workdir 必须在 mount_path 下。
3.7 BoxSkillStore (box/skill_store.py, 647 行)
新增模块（commit 4ab3502），把 skill 持久化收归 Box runtime：
BoxSkillStore
├─ list_skills() / get_skill(name)
├─ create_skill(data) / update_skill(name, data) / delete_skill(name)
├─ scan_skill_directory(path) 扫描目录返回候选 skill 包列表
├─ list_skill_files(name, path) 浏览 skill 内文件树
├─ read_skill_file(name, path) / write_skill_file(name, path, content)
├─ preview_skill_zip(zip_bytes, ...) 不落盘预览 zip 内容
└─ install_skill_zip(zip_bytes, ...) 解压、校验、复制到 skills_root
└─ 支持 source_subdir / target_suffix（commit 1aa043f）
GitHub 安装路径：HTTP 层（api/http/service/skill.py）先 git clone 拉取，再走 install_skill_zip 或 directory 路径。Skill 文件存放于 box.local.skills_root（默认 skills，相对 host_root），容器内对应 /workspace/.skills/。
3.8 Security (box/security.py, 52 行)
validate_sandbox_security(): 黑名单校验 host_path，阻止挂载 /etc//proc//sys//dev//root//boot 及 Docker/Podman socket。
已知缺陷: 根路径 / 未拦截，用户 home 目录未拦截，是 denylist 而非 allowlist 策略。详见 SaaS 阻塞项 S5。
3.9 Errors (box/errors.py, 33 行)
| 异常类型 | 含义 |
|---|---|
BoxError |
基类 |
BoxValidationError |
spec/参数校验失败 |
BoxBackendUnavailableError |
无可用 backend |
BoxRuntimeUnavailableError |
Runtime 服务不可用 |
BoxSessionConflictError |
session 已存在但 spec 不兼容 |
BoxSessionNotFoundError |
session 不存在 |
BoxManagedProcessConflictError |
session 已有同名 process |
BoxManagedProcessNotFoundError |
process 不存在 |
4. 工具系统集成
4.1 ToolManager 编排 (toolmgr.py)
ToolManager.initialize()
├─ NativeToolLoader (exec / read / write / edit / glob / grep)
├─ PluginToolLoader (插件工具)
├─ MCPLoader (MCP Server 工具)
├─ SkillToolLoader (activate 工具 — Tool Call 激活)
└─ SkillAuthoringToolLoader (Skill CRUD)
工具调用优先级: native → plugin → mcp → skill → skill_authoring
4.2 Native Tools (native.py, 846 行)
| 工具 | 是否在 Box 中执行 | 是否访问宿主文件系统 |
|---|---|---|
exec |
是 | 否 |
read |
否 | 是 — 直接 open() 宿主文件 |
write |
否 | 是 — 直接 open() 宿主文件 |
edit |
否 | 是 — 直接 open() 宿主文件 |
glob |
否 | 是 — 直接遍历宿主目录 |
grep |
否 | 是 — 直接读宿主文件 |
沙箱边界不对称: 这是刻意的设计权衡 — read/write/edit/glob/grep 绕过沙箱以获得性能（避免容器 I/O 开销与跨进程拷贝），但意味着 LLM 可以直接读写 allowed_mount_roots 下任何文件。Skill 路径经 _resolve_host_path() 重写，禁止穿越 package_root。
exec 的 Skill 分支: 命令中引用 /workspace/.skills/<name> 的 skill 时：
- 验证 skill 已激活
- 单次 exec 只能引用一个 skill 包
- 若 skill 是 Python 项目（有
requirements.txt或pyproject.toml），命令会被 venv bootstrap 包裹（在 skill 挂载点内创建.venv） - 调用
box_service.execute_tool()→ 走默认 session_id 与已组装好的extra_mounts，不再为每 skill 起独立 session
4.3 MCP-in-Box (mcp_stdio.py, 354 行)
BoxStdioSessionRuntime 让 MCP stdio 服务器在 Box 容器中运行，共享 session、多 process模式（commit 529088e）：
initialize()
1. 复用/创建共享 session (session_id = _build_box_session_id())
- persistent=True，长期保持
2. workspace.execute_raw(install_cmd) 安装依赖 (可选)
3. 将每个 MCP server 文件 stage 到 /workspace/.mcp/<process_id>/
4. workspace.start_managed_process(process_id=<server>)
5. websocket_client(ws_url) 通过 WS relay 连接
6. ClientSession.initialize() MCP 协议握手
配置 (MCPServerBoxConfig): network='on' (MCP 服务器通常需要网络)，host_path_mode='ro' (默认只读)，startup_timeout_sec=120 (留时间给 pip install)。
每条 MCP server 是同一 session 中的一个 managed process，独立的 process_id、独立 attach URL，互不阻塞。
5. 启动与生命周期
5.1 启动顺序 (build_app.py)
BuildAppStage.run(ap)
├─ ... (persistence, models, sessions) ...
│
├─ BoxService(ap)
├─ box_service.initialize()
│ └─ connector.initialize()
│ ├─ [stdio] fork box subprocess
│ ├─ [subprocess+WS] Windows 本地
│ └─ [remote WS] connect URL
│ └─ 启动心跳 _heartbeat_task
├─ ap.box_service = box_service
│
├─ ToolManager(ap)
├─ tool_mgr.initialize()
│ ├─ NativeToolLoader (检查 box_service.available)
│ ├─ PluginToolLoader
│ ├─ MCPLoader (Box 可用时，stdio MCP 走沙箱)
│ └─ SkillAuthoringToolLoader
├─ ap.tool_mgr = tool_mgr
│
├─ ... (platform, pipeline) ...
├─ SkillManager.initialize() (从 Box runtime 加载 skill 列表)
└─ ... (RAG, HTTP, plugins) ...
BoxService 在 ToolManager 之前初始化。ToolManager 创建 loader 时检查 box_service.available。
5.2 初始化失败处理
try:
await self._runtime_connector.initialize()
self._available = True
except Exception as e:
self._available = False
logger.warning(f"Box runtime unavailable: {e}")
静默降级: Box 初始化失败不会阻止应用启动，仅导致 6 个 native tool、所有 Skill 工具和 MCP-in-Box 工具不暴露给 LLM。与 Plugin 的行为不同（Plugin 失败会抛异常）。
5.3 销毁流程
app.dispose()
└─ box_service.dispose()
├─ connector.dispose()
│ ├─ cancel _heartbeat_task
│ ├─ cancel _handler_task / _ctrl_task
│ └─ terminate subprocess (SIGTERM)
└─ loop.create_task(client.shutdown())
└─ RPC SHUTDOWN → Box Runtime 清理所有容器
Box 额外做了 RPC SHUTDOWN 通知 Runtime 主动清理容器，比 Plugin 的直接杀进程更安全。
6. 配置
config.yaml (重构后)
box:
enabled: true # 整个 Box 子系统的总开关。设为 false 时：
# - 不连接远程 Box runtime，不 fork 本地 stdio 子进程
# - sandbox 工具 (exec/read/write/edit/glob/grep) 不暴露给 LLM
# - skill 添加/编辑 / GitHub 安装 / 文件写入全部拒绝
# - stdio 模式的 MCP server 启动时报错（http/sse 模式不受影响）
# - skill 列表/读取保持只读可用
# BOX__ENABLED 环境变量可覆盖（统一约定）
backend: 'local' # 'local' (探测) / 'docker' / 'nsjail' / 'e2b'
# 由 box.backend / BOX__BACKEND 选择后端
runtime:
endpoint: '' # 外部 Runtime 的 WS 基地址 'ws://host:5410'
# 留空 = 本地自管 Runtime
local:
profile: 'default'
image: '' # 覆盖 profile 默认 image
host_root: './data/box' # 工作区挂载根，Docker 部署需绝对路径
default_workspace: '' # 默认 '<host_root>/default'
skills_root: 'skills' # Box 管理的 skill 包目录（相对 host_root）
allowed_mount_roots: # 默认 ['<host_root>']
- './data/box'
- '/tmp'
workspace_quota_mb: null # 配额覆盖，null = 走 profile
e2b:
api_key: '' # 也可走 E2B_API_KEY 环境变量
api_url: '' # 自托管 E2B 时填写
template: '' # 默认 template ID
重大变更: 较 2026-04-16 文档，配置结构完全重组（commit
eefdea4）。原字段box.profile/box.runtime_url/box.shared_host_root/box.allowed_host_mount_roots全部迁入box.local.*子表，新增box.backend与box.e2b.*配置组。
docker-compose.yaml
langbot_box 服务受 compose profile 控制,默认 docker compose up 不会启动它。需要 sandbox 时:
docker compose --profile box up # 启动 langbot + langbot_box + plugin runtime
docker compose --profile all up # 同上
docker compose up # 只起 langbot + plugin runtime (box 关闭)
若不起 langbot_box,需要同步在 data/config.yaml 中设 box.enabled: false(或 langbot 容器 env 加 BOX__ENABLED=false),否则 LangBot 会一直尝试连接不存在的 Box runtime 并报错。
# langbot_box 的关键 volume
volumes:
- ${LANGBOT_BOX_ROOT}:${LANGBOT_BOX_ROOT} # 工作区挂载(源/目标同路径)
- /var/run/docker.sock:/var/run/docker.sock # Docker backend 复用宿主 docker
关闭/连接失败时的行为矩阵
box.enabled = false 与"启用但连接失败"在用户可观察行为上完全一致——都通过 BoxService.available = False 表达,只是 get_status 多返回 enabled 字段供前端区分文案。
| 消费方 | Box 可用 | Box 不可用(disabled 或 failed) |
|---|---|---|
| native exec/read/write/edit/glob/grep 工具 | 暴露给 LLM | 不暴露 |
activate / register_skill 工具 |
暴露给 LLM | 不暴露 |
| stdio MCP server | 在 Box 内启动 | _init_stdio_python_server 抛 RuntimeError 拒绝;不退化到宿主 stdio |
| http/sse MCP server | 正常 | 正常(不依赖 Box) |
Skill 列表/读取 (list_skills/get_skill/read_skill_file) |
走 Box runtime | 走 LangBot 本地 data/skills/ 只读 fallback |
| Skill 创建/编辑/安装/写文件 | 走 Box runtime | HTTP 400 + 明确错误信息(_require_box_for_write) |
Pipeline AI 配置中 box-session-id-template |
正常生效 | 前端 banner 提示字段无效 |
Pipeline 扩展页 enable_all_skills / 绑定 skill |
可编辑 | 前端禁用 + banner |
| 仪表盘 Box 状态卡片 | 绿点 / "已连接" | 灰点 / "已禁用"(disabled) 或 红点 / "已断开"(failed) |
后端拒写的边界条件:如果
ap.box_service完全没装(老式 dev mode,没经过 BuildAppStage),_require_box_for_write视作 no-op,保留data/skills/本地路径——以兼容历史测试与最小化设置。生产环境总会装ap.box_service,因此该 fallback 不会被触发。
Pipeline 配置 (templates/metadata/pipeline/ai.yaml)
local-agent.config.box-session-id-template 控制 session 作用域，预设：
{launcher_type}_{launcher_id}— 每个会话 (推荐，默认){launcher_type}_{launcher_id}_{sender_id}— 群聊每个用户{launcher_type}_{launcher_id}_{conversation_id}— 每个对话上下文{query_id}— 每条消息（完全隔离）
REST API
| 端点 | 方法 | 说明 | 前端 |
|---|---|---|---|
/api/v1/box/status |
GET | 可用性、Profile、后端信息 | ✅ 监控页 |
/api/v1/box/sessions |
GET | 活跃 session 列表 | ❌ |
/api/v1/box/errors |
GET | 最近 50 条错误 | ❌ |
/api/v1/skills 等 |
GET/POST/PUT/DELETE | Skill CRUD、文件浏览、zip/GitHub 安装、preview | ✅ Skill 管理页 |
前端 web/src/app/home/monitoring/components/overview-cards/SystemStatusCards.tsx 已接入 /api/v1/box/status，展示 backend 名称、profile 与活跃 session 数。Sessions 与 errors API 仍未接入。