Replace the sdkTranscript TEXT column with WorkspaceStorageBackend
(GCS/local) for persisting Claude Code JSONL transcripts. This removes
the implicit 512KB cap that caused --resume to degrade after a few
tool-heavy turns (JSONL is append-only and never shrinks).
Key changes:
- Strip progress/metadata entries before storing (~30% size reduction)
with parentUuid reparenting for orphaned children
- Upload in background (asyncio.create_task) to avoid blocking SSE
- Size-based conflict guard: never overwrite a larger (newer) transcript
- Validate stripped content before upload
- Log warning when falling back to compression approach
- Enable claude_agent_use_resume by default
- Remove sdkTranscript column from schema, model, and DB layer
- Storage path: chat-transcripts/{user_id}/{session_id}/{session_id}.jsonl
Capture Claude Code CLI session transcripts via the Stop hook and persist
them in the DB. On subsequent turns, write the transcript to a temp file
and pass --resume so the CLI restores full conversation context without
lossy history compression.
Key changes:
- transcript.py: read/write/validate JSONL transcript utilities
- security_hooks: register Stop hook to capture transcript_path
- service.py: resume strategy with fallback to compression
- schema.prisma: add sdkTranscript column to ChatSession
- Feature flag: CLAUDE_AGENT_USE_RESUME (default off)
