id: local-agent-multimodal-debug-chat
title: "Local Agent Debug Chat preserves uploaded image input"
mode: agent-browser
area: pipeline
type: regression
priority: p2
risk: medium
ci_eligible: false
tags:
  - local-agent
  - multimodal
  - pipeline
skills:
  - langbot-env-setup
  - langbot-testing
env:
  - LANGBOT_FRONTEND_URL
  - LANGBOT_BACKEND_URL
  - LANGBOT_LOCAL_AGENT_PIPELINE_URL
  - LANGBOT_LOCAL_AGENT_PIPELINE_NAME
automation: scripts/e2e/pipeline-debug-chat.mjs
automation_env:
  - LANGBOT_FRONTEND_URL
  - LANGBOT_BROWSER_PROFILE
  - LANGBOT_CHROMIUM_EXECUTABLE
  - LANGBOT_LOCAL_AGENT_PIPELINE_URL
  - LANGBOT_LOCAL_AGENT_PIPELINE_NAME
automation_pipeline_url_env: LANGBOT_LOCAL_AGENT_PIPELINE_URL
automation_pipeline_name_env: LANGBOT_LOCAL_AGENT_PIPELINE_NAME
automation_prompt: "I attached an image. Reply only IMAGE_OK if you received the image."
automation_expected_text: "IMAGE_OK"
automation_image_base64_fixture: "skills/langbot-testing/fixtures/multimodal/red-square.png.base64"
setup_automation:
  - "node:scripts/e2e/ensure-local-agent-pipeline.mjs --write-env"
setup_provides_env:
  - LANGBOT_LOCAL_AGENT_PIPELINE_URL
  - LANGBOT_LOCAL_AGENT_PIPELINE_NAME
preconditions:
  - "The selected model route accepts image input, or the case is intentionally checking graceful provider rejection."
steps:
  - "Prepare a small PNG file for upload. The bundled fixture base64 is at skills/langbot-testing/fixtures/multimodal/red-square.png.base64 if a temporary file is needed."
  - "Open LANGBOT_FRONTEND_URL."
  - "Navigate to Pipelines and open the target local-agent pipeline."
  - "Open Configuration > AI."
  - "Use runner Default or the pluginized langbot/local-agent runner."
  - "Select a model that supports image input in the current environment, or use a known model that at least accepts the uploaded image payload."
  - "Save the pipeline."
  - "Open Debug Chat."
  - "Attach the PNG through the image/file upload control. Prefer the bundled 64x64 red-square fixture; 1x1 images may be rejected by some model providers before runner behavior is exercised."
  - "Confirm the user compose area or sent message shows the image attachment."
  - "Send: I attached an image. Reply only IMAGE_OK if you received the image."
checks:
  - "UI: The sent User message shows an image attachment, not just text."
  - "UI: The Bot message contains IMAGE_OK."
  - "Network or logs: The browser sends an image upload request, or backend logs show the local-agent input contains an image."
  - "Console: No unexpected frontend runtime errors appear during upload or Debug Chat."
evidence_required:
  - ui
  - screenshot
  - console
  - network
  - backend_log
diagnostics:
  - "If the model cannot process image input, repeat with a multimodal-capable model before diagnosing local-agent."
  - "For RAG plus multimodal coverage, keep a KB bound and verify the image remains visible while the answer uses the KB sentinel."
troubleshooting:
  - local-agent-model-route-unavailable
  - plugin-runtime-timeout
  - proxy-env-mismatch
  - provider-image-parse-error
  - survey-widget-blocks-debug-chat