Skip to content

v1.9.2

Choose a tag to compare

@ergut ergut released this 13 Sep 08:39
· 3 commits to main since this release
27fac67

Highlights

  • Search counts no longer leak excluded matches. Total results found and the file list are derived from the filtered result set. A namespace- or tag-restricted client used to see the raw match count with nothing listed, which works as an oracle for whether a term appears in hidden pages, and file paths carried hidden page names. While exclusions are active has_more is always false and the "more results available" hint is never shown, because the API's flag describes the unfiltered set, so results past limit may be cut off without a hint (#102).
  • Deleted pages leave the vector index. The sync walker indexed logseq/bak/ backups and logseq/.recycle/ deleted pages, so a deleted page stayed searchable under a mangled pages/... title and backups polluted results. logseq/ and hidden directories are now skipped; stale entries are dropped on the next logseq-sync run (#102).
  • Scores mean what they say. Hybrid (default) and keyword modes return RRF / BM25 scores where higher is better. They no longer carry distance-based relevance labels or the "lower is more relevant" note, which only applies to vector mode. If a hybrid or keyword search fails and falls back to vector-only, the distance note is shown (#102).

Added

  • list_pages accepts an optional limit so a large graph no longer dumps every page name into the client's context (#102).
  • The MCP initialize response reports the installed package version in serverInfo.version (#102).

Fixed

  • search, query and list_pages accept a limit sent as a JSON number like 20.0 instead of failing with a TypeError (#102).
  • The "Vector DB not initialized" message points at logseq-sync --once instead of the sync_vector_db tool, which does not sync (#102).

Documentation

  • README: tool count, missing get_block row, LOGSEQ_VERIFY_SSL. VECTOR_SEARCH.md: how to read scores, what the sync indexes, vector_search is read-only and sync_vector_db only points at the CLI, log location. DEVELOPMENT.md and TESTING.md: current package layout, how to add a tool with an access policy, stderr logging, regenerated test tree (#102).

Full changelog: https://github.com/ergut/mcp-logseq/blob/main/CHANGELOG.md