1.1.0 - 2026-08-26

Release Notes

Added

• list_topics now returns the directory tree beneath each topic. Every
  entry carries a children array: one node per path segment below the
  category and topic, each with segment, file_count (the documents under
  that prefix, so a parent counts everything beneath it) and its own
  children. Root and category-only entries, and topics whose documents sit
  directly in the topic directory, carry []. The tree is built from
  documents.path, so a document whose frontmatter topic: overrides the
  path-derived topic still contributes the directories after its second path
  segment to the group it was filed under, and siblings are sorted by name so
  the output does not depend on the order SQLite returns rows. This is a field
  addition under the 1.0 freeze (docs/stability.md):
  nothing existing changes shape, and the tool still takes no parameters.

  list_topics_returns_the_directory_tree_beneath_each_group_over_http in
  grooveseek/tests/mcp_protocol_surface.rs reads it back through the
  Streamable HTTP transport from a seeded index, and segment_tree's unit
  tests in grooveseek/src/db/meta.rs pin the rules one at a time — the file
  name is not a node, the first two segments are not repeated, a parent counts
  every document beneath it, siblings are sorted. docs/filters.md also stops
  saying that category can come from a category: frontmatter field; there
  is no such field, only topic:.

• A group that starts with - is excluded from the search. rust -async
  drops every chunk containing async from both halves of the hybrid: the
  full-text half compiles to ("rust") NOT ("async"), and the vector half
  drops the candidates whose chunk matches the same negative expression, so
  one FTS5 judgement (trigram, case-insensitive, diacritics removed) decides
  the question for both legs. The judgment is made against the same FTS row a
  positive match sees — heading, the contextual prefix, and content
  together — not the body alone, so an excluded term in a heading also drops
  the chunk. -"exact phrase" excludes a verbatim phrase; an unquoted
  -word is tokenized with the same rules as the rest of the query, so
  -再ランキング also excludes ランキング — quote it to exclude only the
  compound. The embedder, the reranker and match_spans see the query with
  its exclusions cut out, so a query without one is embedded exactly as
  before. The response echoes what was excluded in
  filter_applied.excluded_terms, and a query made only of exclusions is
  refused ({"error": …} over MCP, stderr and a non-zero exit on the command
  line, a load error for a golden file). An excluded phrase under the
  three-character trigram floor excludes nothing; the parent retriever may
  expand a hit into text that contains the excluded term, since exclusion is
  judged on the hit chunk, not on content a later expansion adds.
  a_chunk_holding_an_excluded_term_never_reaches_the_fts_leg and
  an_excluded_term_drops_the_vector_nearest_chunk_too pin the two halves
  against a real FTS5 table. Rationale in
  ADR-0011.

Changed

• What concurrent HTTP clients pay for the search locks is now measured, and
  the docs say so. docs/clients.md claimed "~10 qps
  expected for search" with nothing behind it. Eight clients at once now have
  a table in docs/deployment-topologies.md,
  taken with cargo test -p grooveseek --release --test http_lock_contention -- --ignored --nocapture:
  search throughput moves from ~7 to ~9 qps on a 9,813-chunk corpus and from
  ~12–16 to ~13–20 qps on a 794-chunk one, latency grows about 4.5× at eight
  clients, and a second daemon on a copy of the same corpus adds only 12–32% —
  one query embedding already runs across every core, so the lock is not
  holding idle hardware back. The database side is where cores wait (the graph
  tool keeps one busy), and its share overtakes the embedding at roughly five
  thousand chunks; below that no lock refactor can raise search throughput.

  grooveseek/tests/http_lock_contention.rs is the instrument: an ignored
  integration test that starts a real groove serve --transport http, releases
  N threads from one barrier against /mcp, and prints the table with the
  three discriminators the decision needed (one embedding versus one hybrid
  fetch timed in-process, two daemons versus one, CPU per request). It asserts
  only that it measured something — non-empty hits, no failed requests, the
  latencies of a round summing to more than the round took — and runs on the
  three-file fixture in the nightly --include-ignored job.

• A - that begins a whitespace-delimited group changed meaning. Until
  now -foo was searched for as the literal token -foo (a hyphen is a word
  character, so sqlite-vec stays one token — that is unchanged). It is now
  an exclusion. To search for a leading hyphen literally, quote it:
  "-foo". ---, a lone -, and - foo are not exclusions. Evaluation
  history is fingerprinted with fts_query_version 3 for this release, so
  groove eval --fail-on-regression will not compare across the change.

Fixed

• The groove eval transcript in the quick start now shows what the binary
  prints, on both pages. docs/eval.ja.md had translated the
  example without its Per-query section, so a Japanese reader was never shown
  the one line that names a query that missed. The English example was not what
  the binary prints either: it lacked the corpus: line every run has carried
  since 0.15.0, ended its per-query row with an expected ... missing phrase
  the formatter has no code for, gave the row a placeholder id where the real
  one is the first 32 characters of the query, wrote the timestamp with a
  +09:00 offset where the binary writes UTC, and its aggregate could not have
  come from its own row — one of two queries at recall@10: 0.00 does not
  average to recall@10 1.000. Both examples are now derived from the golden
  file above them: two hits at ranks 1 and 3 for the first query, a miss for the
  second, recall@10 0.500, MRR 0.500, nDCG@10 0.460. The Japanese page
  also gains the sentence on heading-less expected hits and the expansion of
  nDCG that its English twin already had.

  grooveseek/tests/docs_eval_transcript.rs keeps it that way. It parses the
  golden fence on each page, runs the metric and formatting code the binary
  runs over the hits the example assumes, and requires the transcript fence to
  equal the output character for character. The ranks are the example's
  premise, not a measurement — what the test pins is that the numbers, the
  layout and the row id follow from them. eval::query_id is public for it,
  so the row id rule lives in one place.

• The MCP tool descriptions, and twenty-three sentences that still credited
  rmcp with checks it had stopped performing. ADR-0009
  moved Host and Origin validation into groove, which hands rmcp empty
  allow-lists so it matches nothing; the doc comments around that code, the
  anyhow::bail! an operator sees for an allowed_origins entry that will not
  parse, and CONTRIBUTING.md's clippy step had not all been told. The two
  that reach a caller are the description= strings, which are all an LLM
  client is given: rebuild_index refuses a call that arrives during a rebuild
  and did not say so, and get_connection_graph returns snippet on every
  node plus truncated and truncation[] on the envelope and named neither.
  Both facts were already on docs/mcp-tools.md and missing
  from the only string a client reads.

  tools/list now carries an assertion about description content in
  grooveseek/tests/mcp_protocol_surface.rs, honest about being a substring
  check: it catches a fact being deleted, not a description drifting from the
  page.

Removed

• The .kb-mcpignore migration check, and the module behind it. v1.0.0
  added two groove doctor findings about an ignore file left under the name
  the project used before
  ADR-0007 — and
  said in the same breath that they were for the migration and would go in
  1.1.0. The module doc, the docs/ARCHITECTURE.md row and
  docs/usage.md all carried that date. This is that removal:
  grooveseek/src/legacy.rs, the findings it fed
  (indexed-despite-legacy-ignore and legacy-ignore-not-examined), and
  ExclusionRules::ignore_only_from_bytes, which had no other caller, are all
  gone. A .kb-mcpignore still keeps nothing out — that is ADR-0007's
  decision and has not changed — but doctor no longer opens one to say so.

  doctor therefore asks two groups of question rather than three, and
  doctor::run takes (db, registry) rather than also a kb_path and an
  exclude_dirs. That signature is internal: groove doctor still requires
  --kb-path to find the index, and every remaining check, exit code and JSON
  field keeps its name, type and meaning. The tests that covered the removed
  findings went with them — the unit tests in src/doctor.rs, two in
  tests/doctor_cli.rs (one of them the #[ignore] end-to-end run that
  proved the remedy) and the four helpers only those two called.

Internal

• Four comments named something other than what the code does. feature-55
  demoted build_fts_query to a #[cfg(test)] helper, and two comments
  outside its file still described it as the entry point production calls: the
  whole-query fallback is assembled by parse_query into a field
  query_phrases does not return, and the round-trip counter a db.rs test
  pins sits above the early return where ParsedQuery::match_expr gives
  None. Two more wrote QueryDiagnostics::idf_clamped in plain backticks
  where the paragraph beside them already linked it.

  Which of those names could become intra-doc links was settled by running
  rustdoc rather than by reading, because whether a link resolves depends on
  the module doing the linking and not on the item alone:
  crate::server::search::MATCH_SPAN_MAX_TERMS resolves from server.rs,
  whose own private mod search it is, and not from db/search.rs.
  fallback_whole_query resolves from nowhere outside db, so it stays a
  backtick with its file named in prose. Checking for a generated rustdoc page
  instead answers a different question and gets both of those backwards.

• The CI command block is now compared with the workflow it copies.
  AGENTS.md and CONTRIBUTING.md each carry the commands that reproduce
  CI, and #229 pinned the two copies to each other -- which left the thing
  they are copies of, .github/workflows/ci.yml, outside the comparison:
  both could drift from the workflow together and agree with each other the
  whole way down. grooveseek/tests/docs_commands_pinned.rs now reads the
  workflow's run: steps through the same reader as the fenced block and
  requires the two sets of commands to be equal, naming the command and the
  side -- jobs.clippy.steps[3], or the block -- that has it alone. Order
  between jobs is not compared: the block is written cheapest-first for
  someone reproducing CI by hand, and the workflow's jobs run in parallel, so
  neither order is the other's. Order within a job is: cargo test --test index_progress_cli runs before cargo test so that one process warms the
  model cache, and a block that lists them the other way round sends a
  reader into the race that order avoids. A step the reader cannot classify, a run: that reads as no
  command, and a line inside a run: the reader cannot place are each
  reported rather than skipped, since any of them is a command CI may run
  that is being compared with nothing. Only run: steps are read: every
  uses: in the workflow today is setup, and the module doc names a check
  written as an action as the thing this cannot see.

  The workflow reader lives in grooveseek/tests/common/workflow.rs, and
  grooveseek/tests/bench_targets_run_in_ci.rs now reads nightly.yml
  through it instead of scanning the file as text -- so a --bench name that
  survives only in a comment no longer counts as run, while a line the reader
  cannot place is still scanned as text rather than dropped. Two private
  copies of the repository-root helper, in that test and in
  diagnostics_stay_ascii.rs, are folded into the shared one.

• Every line of a shell block in the documentation is now accounted for.
  The reader behind the command-copy guards (#229) was checked one block at a
  time: a fenced shell block that yielded no command failed the suite, but a
  block that yielded one command and dropped the rest passed, and because both
  translations dropped the same lines, the guards comparing them compared
  nothing there and stayed quiet. common::docs::read_block now returns every
  raw line of a block with what the reader made of it -- an instruction, a
  continuation, a heredoc payload, a blank, a comment, grammar, or unread with
  the reason -- and command_lines is a filter over it, so the function that
  decides a line is not a command is the one that says why. A new guard in
  docs_commands_subset.rs fails on any line left unread, naming the page,
  the line and the reason.

  Written against today's tree, the guard named thirteen lines in three
  shapes, and the reader learned each: a quoted argument that closes lines
  later (python -c " in the Windows quirks skill, jq ' in the full-audit
  command -- the payload is now part of the instruction, and the && mv
  chained after jq's closing quote is compared for the first time), the arms
  of a case (the pattern is a branch condition and ;; is grammar; what
  stands between them is read), and a PowerShell assignment ($action = New-ScheduledTaskAction ... keeps the cmdlet, as NEXT_ID=$(jq ...) keeps
  jq). Before this, a payload line of that python -c block was being read
  as a shell assignment.

  What never closes is reported rather than swallowed: a quote open at the end
  of a block, a heredoc without its terminator, a \ on the last line -- each
  of which used to take every line under it into silence. Four things the old
  reader did are gone, none of which any page relied on: a \ at the end of a
  comment continued the line, a << inside a comment opened a heredoc, a \
  inside an open quote was a continuation, and that last-line \ vanished.
  The block-against-block count the subset guard's header quotes moved from
  fifteen to sixteen: the jq block now names mv, and the block above it,
  which runs jq alone, is a subset of it.

• The source layout table in docs/ARCHITECTURE.md
  is now compared with the tree. The table went unrepaired across #195,
  #196 and #197, which each split a module out of server.rs, and #212,
  which added legacy.rs; a docs-only #214 repaired it by hand, because
  nothing read it. grooveseek/tests/docs_source_layout.rs now walks every
  file under each workspace member's src and requires both tables to
  describe it: a row of its own, or its directory's row naming it in
  backticks inside that row — the rule #214 worked out by hand, since a match
  by file name hides server/search.rs behind db/search.rs — or, for a
  member the table describes with a single crate-level row such as
  crates/groove-svc/, that row alone. The other
  direction too, compared as written so a wrong case fails on Windows as it
  does on Linux; the English and Japanese tables in the same order; and a row
  the reader cannot classify, or one with a | the parser would swallow, is
  reported rather than skipped. The pages as they stood before #214 are
  frozen under grooveseek/tests/fixtures/docs-history/, and the test
  requires exactly those four modules to be what it reports there.

  The main.rs row no longer copies the subcommand list. README and
  docs/index carry it under a sentence the binary's documented_flags
  module already compares with Cli::command(), and a second copy in a table
  row is what the table just spent four pull requests demonstrating. It
  points at docs/usage.md instead.

• A sentence that says it lists every command is now compared with the
  binary. README.md, README.ja.md and the two docs/index pages each
  open their reference table with "Every command: index, serve, …", and
  the README once shipped that sentence with status missing — nine names
  under a heading that promises ten. The flag guards could not see it: they
  read tokens that begin with --, and a subcommand name has no such shape.
  A test in the binary's documented_flags module now reads every such
  sentence and requires the enumerated set to equal Cli::command()'s
  subcommands, nothing missing and nothing extra; the same test pins
  groove service <verb> in both directions against
  docs/usage.md.

• A command list copied into a second place now fails the test suite when the
  copies disagree. Three guards under grooveseek/tests/ read every Markdown
  page the repository publishes. docs_commands_subset.rs reports a page that
  tells a reader to run a shortened copy of a command block it already carries;
  docs_commands_pinned.rs holds the snippets published in more than one place
  on purpose — the CI commands in AGENTS.md and
  CONTRIBUTING.md, the minimal .mcp.json in the README and
  docs/clients.md, the hook recipe in
  grooveseek/examples/hooks/ — each marked with an
  invisible <!-- groove-pin: id --> and compared as a value rather than as
  characters; and docs_commands_twins.rs requires an English page and its
  Japanese counterpart to name the same commands.

  The last of those found what it was built for.
  CONTRIBUTING.ja.md still gave Japanese readers a
  four-command line to run before opening a pull request — the shortened copy
  the English page deleted in #228, naming one clippy leg where CI runs two.
  The block it was copied from is on the same page and says so outright, in a
  different section eighty lines up, which is exactly how a copy drifts without
  either half looking wrong. It now points at that block, as the English page
  does.

  Commands compare by identity — program plus subcommand — and not by their
  flags, because a copy drifts in its flags first. The subset rule is
  one-directional for a measured reason: two fenced blocks on one page stand in
  a subset relation fifteen times here today, in AGENTS.md, docs/usage.md
  and two deployment READMEs, and every one is deliberate, so a symmetric rule
  would report fifteen false alarms on a clean tree.

grooveseek 1.1.0

Download grooveseek 1.1.0

File	Platform	Checksum
grooveseek-aarch64-apple-darwin.tar.xz	Apple Silicon macOS	checksum
grooveseek-x86_64-pc-windows-msvc.zip	x64 Windows	checksum
grooveseek-aarch64-unknown-linux-gnu.tar.xz	ARM64 Linux	checksum
grooveseek-x86_64-unknown-linux-gnu.tar.xz	x64 Linux	checksum

groove-tray 1.1.0

Download groove-tray 1.1.0

File	Platform	Checksum
groove-tray-x86_64-pc-windows-msvc.zip	x64 Windows	checksum

groove-svc 1.1.0

Download groove-svc 1.1.0

File	Platform	Checksum
groove-svc-x86_64-pc-windows-msvc.zip	x64 Windows	checksum