feat(telemetry): payload v2 with feature usage counters and instance heartbeat

Per-query events now carry event_type='query' and a features JSON object:
- tool_calls by source (native/plugin/mcp/skill) via ToolManager
- tool_call_rounds, kb usage (count/engine plugins/retrieved entries) via local-agent
- sandbox execs/errors via BoxService
- activated_skills and bound mcp_servers snapshots

New instance_heartbeat event (startup + daily) reports anonymous instance
profile: deploy platform, database/vdb kind, box backend/availability,
adapter type names, and resource counts. Respects space.disable_telemetry.

All collection helpers are defensive and never break the pipeline.
Verified: ruff, 37 telemetry unit tests (13 new), 504 box/provider/pipeline tests.
This commit is contained in:
RockChinQ
2026-06-12 08:11:43 -04:00
parent bca710dbd4
commit dd96da895c
10 changed files with 488 additions and 0 deletions
+11
View File
@@ -68,10 +68,21 @@ class TelemetryManager:
'edition',
'error',
'timestamp',
'event_type',
):
if sfield not in sanitized:
continue
v = sanitized.get(sfield)
sanitized[sfield] = '' if v is None else str(v)
# event_type defaults to 'query' for backward compatibility
if not sanitized.get('event_type'):
sanitized['event_type'] = 'query'
# features must be a JSON object
if 'features' in sanitized and not isinstance(sanitized['features'], dict):
sanitized['features'] = {}
if 'duration_ms' in sanitized:
try:
sanitized['duration_ms'] = (