This release includes breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
+6 more
ReleasePort's take
Moderate signalThe `reanalyze` command is deprecated; use `longhand analyze --all`. It will be removed in v1.0.
Why it matters: Deprecation severity 70 triggers removal of `reanalyze` in version v1.0; migrate immediately to avoid breakage.
Summary
AI summaryCLI help is grouped, stats are split into low‑confidence episodes with accurate resolved rates, and first‑run output uses real data.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Medium |
`analyze --all` now rebuilds episode embeddings, absorbing `reanalyze` functionality. `analyze --all` now rebuilds episode embeddings, absorbing `reanalyze` functionality. Source: llm_adapter@2026-07-15 Confidence: high |
— |
| Feature | Medium |
`longhand stats` splits out `low_confidence_episodes` and reports `resolved_rate_pct` over substantive episodes only. `longhand stats` splits out `low_confidence_episodes` and reports `resolved_rate_pct` over substantive episodes only. Source: llm_adapter@2026-07-15 Confidence: high |
— |
| Feature | Medium |
First‑run `setup` ends with an indexed‑summary line and suggests a project‑specific status command. First‑run `setup` ends with an indexed‑summary line and suggests a project‑specific status command. Source: llm_adapter@2026-07-15 Confidence: high |
— |
| Deprecation | High |
Deprecates `longhand reanalyze`; use `longhand analyze --all`. Will be removed in v1.0. Deprecates `longhand reanalyze`; use `longhand analyze --all`. Will be removed in v1.0. Source: llm_adapter@2026-07-15 Confidence: high |
— |
| Refactor | Low |
`longhand --help` groups commands into six panels. `longhand --help` groups commands into six panels. Source: llm_adapter@2026-07-15 Confidence: high |
— |
| Refactor | Low |
Plumbing commands are hidden from default listings. Plumbing commands are hidden from default listings. Source: llm_adapter@2026-07-15 Confidence: high |
— |
Full changelog
UX release: the CLI gets organized, the stats get honest, and first-run gets smarter. No breaking changes — every existing command and flag still works.
Changed
longhand --helpis now grouped. Commands render in six panels — Recall, Archaeology, Browse & insights, Data, Setup & health, Plumbing — instead of a 40-command alphabetical list. (#28)- Plumbing commands are hidden (still fully callable, exact names unchanged):
ingest-session,ingest-live,context,backfill-episodes,mcp-server. These are hook/desktop-config entry points, not user commands. (#28) analyze --allnow also rebuilds episode embeddings, absorbing whatreanalyzedid.reanalyzeremains as a hidden deprecated alias that warns and delegates — it will be removed in v1.0. (#28)- Honest episode stats.
longhand statsand theget_statsMCP tool now split outlow_confidence_episodes(fixless extractions below 0.5 confidence — probes and tool churn, not real problems) and reportresolved_rate_pctover substantive episodes only. Your resolved rate was never as bad as the old denominator made it look. (#28) - First-run output teaches with your own data.
setupends with an indexed-summary line (sessions · projects · episodes) and suggestslonghand status "<your top project>"instead of a placeholder. (#28)
Deprecated
longhand reanalyze→ uselonghand analyze --all. Alias removed in v1.0.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Wynelson94/longhand
Persistent local memory for Claude Code. Indexes every session JSONL verbatim into SQLite + ChromaDB for semantic recall (~126ms)
Related context
Related tools
Earlier breaking changes
- v0.13.0 Deprecates several CLI commands; they will be removed at v1.0.
Beta — feedback welcome: [email protected]