doobidoo/mcp-memory-service v10.57.0

v10.57.0 — AND/OR Tag Filtering for memory_list + Automatic Session Chunking

on GitHub

Special Thanks

Huge thanks to @filhocf for both feature PRs in this release — continuing the excellent streak of quality contributions!

What's New

AND/OR Tag Filtering for memory_list (PR #904)

memory_list now accepts a tag_match parameter, harmonizing it with memory_search and memory_delete:

• tag_match: "any" (OR, default — existing behavior unchanged): returns memories matching at least one of the supplied tags
• tag_match: "all" (AND): returns only memories matching every supplied tag

Implemented across all three storage backends (sqlite_vec, cloudflare, hybrid) and the REST API.

Example:

{
  "tags": ["python", "architecture"],
  "tag_match": "all",
  "page_size": 20
}

Automatic Session Chunking for memory_store_session (PR #912)

Long sessions stored via memory_store_session are now automatically split at turn boundaries (User:, Assistant:, Human:, AI:, You:, Bot:). Each chunk gets its own embedding for better semantic retrieval.

• SESSION_CHUNK_SIZE env var (default: 1500 chars, set to 0 to disable)
• Chunks tagged chunk:N/M for ordered sequential retrieval
• Fail-fast: if any chunk fails to store, the operation halts immediately
• Fully backward compatible — short sessions unchanged

Example chunk tags: ["session:abc-123", "chunk:1/3"], ["session:abc-123", "chunk:2/3"], ...

Bug Fixes

• CI: guard pr-contributor-welcome against non-PR event triggers (commit b42cf00): Added github.event_name == 'pull_request_target' guard to prevent workflow crash on push events.

Upgrade

pip install --upgrade mcp-memory-service
# or
uv pip install --upgrade mcp-memory-service

No breaking changes. No database migration needed.

Links

• CHANGELOG
• Full diff v10.56.3...v10.57.0