docs: epic D PR body for review

Author: Shahinyanm

.docs/plans/2026-05-07-v0.2.1-epic-d-pr-body.md

@@ -0,0 +1,61 @@
+## Summary
+
+Epic D — v0.2.1 operational maturity. **9 atomic commits** on `claude/v0.2.1-epic-d`, built off `claude/v0.2.0-epic-c` HEAD. **Non-breaking** — minor bump after 0.2.0.
+
+> **Merge order:** epic A → main, epic B (rebased) → main, epic C (rebased) → main → tag `v0.2.0`. Then this branch (rebased) → main → tag `v0.2.1`.
+
+Plan: [`.docs/plans/2026-05-07-v0.2.1-epic-d-operational.md`](./.docs/plans/2026-05-07-v0.2.1-epic-d-operational.md)
+
+### What changed
+
+**Performance**
+- `perf(mcp)` — process-wide `Arc<Mutex<rusqlite::Connection>>` cache keyed by state path. First call opens; later calls reuse. Eliminates per-call PRAGMA + migrations replay.
+
+**User-facing DX**
+- `feat(export)` — `task-journal export --format sqlite` produces a clean VACUUM-based snapshot, streamable to stdout for `> backup.sqlite`. Round-trips through `task-journal pack` from a fresh XDG.
+- `feat(cli)` — `task-journal pending list` and `pending retry`. Surface auto-capture-hook failures that used to sit silently in `pending/`. `attempts` counter; rename to `<id>.dead.json` after 3 failures so they stop being retried but still appear in `list`.
+
+**Observability**
+- `feat(mcp)` — structured tracing with `correlation_id` per tool call. Two INFO log lines (start + ok / err) wrap each handler. Default `RUST_LOG=info` gives one greppable line per request.
+- `feat(mcp)` — graceful Ctrl-C / SIGTERM (Unix only) shutdown via `tokio::select!` between rmcp serve loop and `wait_for_shutdown_signal()`.
+
+**Quality**
+- `test(mcp)` — rmcp client + transport compile-and-shape integration test. Full E2E roundtrip deferred to follow-up `claude-memory-yj1.8` (needs `TaskJournalServer` extracted into a lib target — out of scope for D).
+
+**Release**
+- `release` — workspace version 0.2.0-rc.1 → 0.2.1; CHANGELOG entry.
+
+### Verification
+
+- `cargo fmt --all -- --check` ✅
+- `cargo clippy --workspace --all-targets -- -D warnings` ✅
+- `cargo test --workspace --all-targets` ✅ — **213 tests** (was 202 from epic C; +11 added by this PR)
+- `cargo bench --workspace --no-run` ✅
+- `cargo build --workspace --release` ✅ — 0.2.1 binaries
+
+### New CLI surface
+
+| Command | Purpose |
+|---------|---------|
+| `task-journal export --format sqlite` | VACUUM-based clean SQLite snapshot to stdout |
+| `task-journal pending list` | List queued classifier failures |
+| `task-journal pending retry [--mock-*]` | Re-feed pending entries; mark dead after 3 |
+
+### New env vars
+
+None beyond what the existing `RUST_LOG` already controls. The structured-tracing output is tied to it.
+
+### Test plan
+
+- [ ] Branch CI green on three OS (`test`, `msrv`, `audit`, `benches-compile`, `coverage`).
+- [ ] Smoke run `task-journal-mcp` from a real MCP client; observe `tool_call start/ok` lines in stderr; SIGTERM exits 0 within ~1s.
+- [ ] `task-journal export --format sqlite > backup.sqlite` then `sqlite3 backup.sqlite '.schema'` shows the v001+v002 tables.
+- [ ] After dogfooding 0.2.0 + this branch for ~3 days, tag `v0.2.1` and `cargo publish` (after rebasing on main once epics A/B/C are landed).
+
+### Out of scope / deferred
+
+- `claude-memory-yj1.8` — extract `TaskJournalServer` into a `tj-mcp` library target. Unblocks the full E2E rmcp roundtrip test we deferred from D4. Tracked as a side-quest for a future epic.
+- Telemetry endpoint (still requires hosted backend).
+- `task-journal compact` (lifecycle archival of closed tasks) — wants a design pass, deferred.
+
+🤖 Generated with [Claude Code](https://claude.com/claude-code)