Skip to content

Wynelson94/longhand

v0.9.1 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 15d MCP Data & Storage
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai-memory anthropic chromadb claude-code forensic local-first
+6 more
lossless mcp mcp-server python recall sqlite

Summary

AI summary

All 19 MCP tools now include title, annotations, and outputSchema for better transparency.

Changes in this release

Feature Low

Adds title, annotations, and outputSchema to all 19 MCP tools.

Adds title, annotations, and outputSchema to all 19 MCP tools.

Source: llm_adapter@2026-07-15

Confidence: high

Feature Low

Introduces helper constants _READ_ONLY and _RECONCILE_HINTS in mcp_server.py.

Introduces helper constants _READ_ONLY and _RECONCILE_HINTS in mcp_server.py.

Source: llm_adapter@2026-07-15

Confidence: high

Feature Low

Adds anti-pattern callouts to descriptions of 14 MCP tools.

Adds anti-pattern callouts to descriptions of 14 MCP tools.

Source: llm_adapter@2026-07-15

Confidence: high

Feature Low

Tightens server.json description to mention tool count and capability grouping.

Tightens server.json description to mention tool count and capability grouping.

Source: llm_adapter@2026-07-15

Confidence: high

Bugfix Low

States no behavior changes and no breaking schema changes in this release.

States no behavior changes and no breaking schema changes in this release.

Source: llm_adapter@2026-07-15

Confidence: low

Full changelog

Tool definition quality pass. Every MCP tool now serializes a human-readable
title, an annotations block (readOnlyHint, openWorldHint, plus
idempotentHint on reconcile), and — for the 17 tools whose response is
pure structured JSON — an outputSchema describing the response shape.

Added

  • title, annotations, and outputSchema on all 19 MCP tools. Two
    helper constants (_READ_ONLY and _RECONCILE_HINTS) at the top of
    mcp_server.py keep the per-tool blocks readable.
  • Anti-pattern callouts in the descriptions of the four lowest-scoring
    tools (get_stats, replay_file, get_project_timeline,
    get_session_commits) and the borderline tools (find_episodes,
    get_episode, find_commits, list_projects, list_sessions,
    get_file_history, replay_file) — every tool description now names
    what NOT to use it for, making tool selection sharper.
  • server.json description tightened to mention tool count and capability
    grouping.

Why

GLAMA scores MCP tool definitions on six dimensions, with outputSchema
and annotations being two MCP-spec fields most directly tied to
"Behavioral Transparency" and "Contextual Completeness." This release adds
both across the surface so MCP clients (and ranking sites like GLAMA) can
treat each tool as documented metadata rather than free-form prose.

No behavior changes. No breaking schema changes.


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

Track Wynelson94/longhand

Get notified when new releases ship.

Sign up free

About Wynelson94/longhand

Persistent local memory for Claude Code. Indexes every session JSONL verbatim into SQLite + ChromaDB for semantic recall (~126ms)

All releases →

Related context

Earlier breaking changes

  • v0.13.0 Deprecates several CLI commands; they will be removed at v1.0.

Beta — feedback welcome: [email protected]