feat(skills): durable reflection-run record with activity timeline surface

[jamiepine]

What

Adds a persisted reflection_runs table and event for the skill-reflection pump, closing the remaining user-visible part of the shipped reflection loop. Every reflection pass now records its lifecycle and surfaces in the portal timeline.

Changes

Database: New reflection_runs migration with agent/channel identity, trigger source (turn_work / worker_success / reflection), referenced worker IDs (JSON), start/end timestamps, terminal status (success / no_op / error / cancelled), declared rationale (separate from observed actions), outcome summary, affected skill identifiers, and token usage slot.

Event: ProcessEvent::ReflectionRunCompleted on the shared event bus — same pattern as BranchResult, WorkerComplete, etc. — piped through the existing ApiEvent SSE pipeline. ApiEvent::ReflectionRunCompleted with the reflection_run_completed SSE event type.

Persistence: ReflectionRunLogger in conversation/history.rs, fire-and-forget matching ProcessRunLogger. Write-on-start (INSERT), guarded completion update (UPDATE … WHERE status = 'running'), preventing duplicate completions.

UI: InlineReflectionRunCard in PortalTimeline.tsx — a quiet divider bar showing "Learned: …" / "No change: …" / "Error: …" — same inline pattern as chronicle checkpoints, no parallel event system.

Validation

• cargo fmt --check — clean
• cargo check — clean
• cargo test --lib — 594 passed, 0 failed
• cargo test --test '*' — 117 passed, 0 failed
• 12 new unit tests covering: no-op detection, error detection, success with skill extraction, duplicate avoidance, keyword exclusion, default signal state
• bun run build — clean
• just preflight — passed
• just gate-pr — all gates green (fmt, check, migration conflicts, lib tests, integration tests)

Intentional deferments (documented in skill-lifecycle.md)

• Curation (Phase 4): deterministic stale/archive pass, LLM consolidation, snapshots + rollback
• Full surfaces (Phase 5): POST /agents/skills/write, CreateSkill.tsx, full SkillInspector usage row, write_approval staging mode
• CLI: pin/adopt/archive/restore commands
• Token usage: slot exists on the row but is not populated yet (requires branch-level token accounting)
• Observed actions: slot exists but is populated with empty array — tool-call-level observability for reflection branches is future work

Design invariants preserved

• Reflection remains silent in conversation (timeline-only)
• Agent-origin skill mutation rails unchanged (WriteOrigin::Agent, read-before-write, archive-on-delete)
• Cron conversations never reflect (ChannelKind::suppresses_reflection)
• No tool authority expansion
• No hidden chain-of-thought stored
• Duplicate completions guarded by WHERE status = 'running'

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 253afe2d-ddbb-4d9c-ac37-978245294155

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}