boost-skill-cli 1.2.131__tar.gz → 1.2.133__tar.gz

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/.lighthouserc.json

@@ -28,7 +28,7 @@
           "matchingUrlPattern": "roadmap\\.html$",
           "aggregationMethod": "median-run",
           "assertions": {
-            "categories:performance": ["error", { "minScore": 0.8 }],
+            "categories:performance": ["error", { "minScore": 0.75 }],
             "categories:accessibility": ["error", { "minScore": 0.95 }],
             "categories:best-practices": ["error", { "minScore": 0.9 }],
             "categories:seo": ["error", { "minScore": 0.95 }]

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/CLAUDE.md

@@ -63,31 +63,34 @@ corpus: `scripts/ensure_eval_corpus.sh` first taps the pinned repo list in
 --defaults` is NOT enough, it omits every rule/workflow repo). The list is
 **twenty** repos: the first six cover every golden target, the rest exist so the
 corpus is a realistic size. That matters more than it sounds — over the six
-alone (743 entries) BM25 scores 0.978 / 0.791 / 0.854 / 0.882, and over the
-twenty (10,152) it scores **0.852 / 0.473 / 0.605 / 0.657** on the same golden
-set, so three of the four old floors fail once the corpus stops being tiny.
+alone (921 entries at the current pins) BM25 scores 0.989 / 0.769 / 0.848 /
+0.880, and over the twenty (10,731) it scores **0.841 / 0.484 / 0.607 / 0.655**
+on the same golden set (what `tests/eval/baseline.json` records), so three of
+the four old floors fail once the corpus stops being tiny.
 
 **The ranked list de-duplicates on the content hash, not the name.** A grade key
 decides both relevance and identity, and keying identity on the name collapsed
 13 different skills called `code-reviewer` into one rank slot — crediting the
 ranker with a compression that existed only in the scoring code, and worth about
 one query of recall@10. That is where the old "recall is 1.000" folklore came
-from; the six-repo corpus measures 0.978 once mirrors collapse and homonyms do
-not. Relevance is still decided by name (or by content class when a golden row
-pins an `exemplar`), so the sets can migrate a row at a time.
+from; the six-repo corpus measured 0.978 at the pins of the time once mirrors
+collapse and homonyms do not. Relevance is still decided by name (or by content
+class when a golden row pins an `exemplar`), so the sets can migrate a row at a
+time.
 
 Each floor sits ~10% under its measured value — loose enough that upstream drift
 can't flake the build, tight enough to catch a collapse. Regression-vs-baseline
 stays relaxed (`--regression-eps 1`), so the absolute floors are the real gate.
 
 **Every row of `taps.txt` pins a commit SHA**, and `scripts/eval_corpus.py`
-checks each clone out at it — the corpus is 10,152 entries, of which one
-third-party repo is 6,309 (62%), so an unpinned list left a required check
+checks each clone out at it — the corpus is 10,731 entries, of which one
+third-party repo is 6,634 (62%), so an unpinned list left a required check
 hostage to someone else's push. CI and `make eval` must invoke the gate with
-identical floors; `tests/unit/test_eval_corpus.py` fails the build if they
-diverge, which they had (CI floored recall alone at 0.85 against a measured
-0.863 — a buffer of 1.15 queries out of 91 — and applied none of the other
-three). Moving a pin means regenerating the baseline; the file says how.
+identical arguments, not just identical floors — `-k` decides what two of them
+measure; `tests/unit/test_eval_corpus.py` parses both calls and fails the build
+if they diverge, which they had (CI floored recall alone at 0.85 against a
+measured 0.863 — a buffer of 1.15 queries out of 91 — and applied none of the
+other three). Moving a pin means regenerating the baseline; the file says how.
 
 **Baselines are keyed by query set** (`name@content-digest`), so one file holds
 both `golden.jsonl` and `golden-natural.jsonl` without either overwriting the
@@ -116,7 +119,20 @@ opt-in evals stay out of `check` and all degrade cleanly:
   avoid. Rates carry a Wilson interval over N runs and floors are judged
   against the bound, not the point estimate: at 3/3 the normal approximation
   claims [1.00, 1.00] and would let a wording regression hide behind one lucky
-  run. Drives a real host, so it is opt-in, out of `check`, and degrades
+  run. A bound judged that way needs **enough N to be reachable**, and the
+  ceiling half is where that bites: at k=0 the Wilson upper bound is
+  z²/(n+z²), a function of N alone, so at the old `--runs 1` default a
+  flawless host was told it FAILED (0.3244 against a 0.20 ceiling, which no
+  sample below n=16 can clear). `--runs` defaults to 3, `min_n_for_ceiling` computes the
+  minimum for any ceiling, and a sample below it reports INCONCLUSIVE with its
+  own exit code (2) rather than a red that says nothing about the host —
+  unless its Wilson *lower* bound is already over the ceiling. Too small to
+  pass is not too small to fail: a conclusive red below the minimum is still a
+  red (exit 1), and a no-call half with no observations at all is
+  INCONCLUSIVE, never a pass. The
+  ceiling is 0.25 so one slip in 24 passes and two do not — a zero-tolerance
+  half beside a floor that absorbs four misses is not the same measurement
+  twice. Drives a real host, so it is opt-in, out of `check`, and degrades
   cleanly when `claude` is not on `PATH`.
 
 The `[eval]` extra (`pip install -e '.[eval]'`) carries `ranx` + `ragas`; nothing
@@ -295,8 +311,11 @@ line coverage. Target `boost_cli/core` behavior with assertions, not just import
   `noxfile.py` opens with `# Copyright the boost contributors.` and
   `# SPDX-License-Identifier: Apache-2.0`. Run
   `python3 scripts/add_spdx_headers.py` (idempotent) — the file list and the
-  expression live there, so changing the licence expression is one edit rather
-  than hundreds. Separately, `prose-lint.yml` names the Markdown files vale checks
+  expression live there (`SPDX_ID`), so changing the licence is one edit
+  rather than hundreds, and `tests/unit/test_spdx_headers.py` pins this
+  paragraph to that constant: #587 moved the tree off `GPL-3.0-only` and left
+  this sentence naming the old expression for every agent that read it since.
+  Separately, `prose-lint.yml` names the Markdown files vale checks
   **explicitly**: a new doc that is not added to that list is never linted, and
   `make lint` will not tell you.
 
@@ -608,9 +627,15 @@ upgrade rather than the entry fee. When both engines are built, `retrieve_any`
 fuses them with reciprocal rank fusion (`rag.rrf_fuse`, `RRF_K = 60`) and
 reports `hybrid RRF`; it degrades to whichever single engine is ready, and to
 BM25 alone otherwise. `dense.status()` names which of the three links (extra,
-backend, built store) is missing, and `dense.fix_hint()` maps that to the one
-next action — both `boost doctor` and `boost search` read that same table, so
-they can't give contradictory advice.
+backend, built store) is missing — or `disabled`, a fourth state that is no
+missing link at all but the `BOOST_NO_EMBED` kill switch, which sits **first**
+in the ladder (`embed.provider()` reads it before any key, so every other
+remedy is a measured no-op under it) and is excluded from `degraded`, so a
+deliberate opt-out never moves doctor's exit code. `dense.fix_hint()` maps
+whichever state it is to the one next action — `boost doctor`, `boost search`,
+the MCP `SEARCH ENGINE` line and both shard surfaces (`reindex --fetch-shards`,
+`update --shards`) all read that same table, so they can't give contradictory
+advice.
 
 **The dense store ranks twice, and `vec0` is why.** `sqlite-vec` has no ANN
 index: a float32 `MATCH` scores *every* vector in the store, which on a real

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/Makefile

@@ -122,15 +122,21 @@ dist-check:
 # 0.000 passed.
 #
 # The floors are calibrated against a TWENTY-tap corpus, not the six it used to
-# be. Over six repos BM25 scored 1.000 / 0.791 / 0.854 / 0.882; over twenty it
-# scores 0.863 / 0.473 / 0.607 / 0.662 on the same golden set. The second set is
-# what a real user sees, so flooring against the first was measuring the corpus
-# rather than the retrieval — three of the four old floors fail outright once
-# the corpus is realistic. Each floor now sits ~10% below its measured value:
-# loose enough that upstream repo drift cannot flake the build, tight enough
-# that a collapse fails it. Regression-vs-baseline stays relaxed
-# (--regression-eps 1) because the corpus tracks upstream HEAD rather than
-# pinned commits; the absolute floors are the real gate.
+# be. Over the six (921 entries) BM25 scores 0.989 / 0.769 / 0.848 / 0.880;
+# over twenty it scores 0.841 / 0.484 / 0.607 / 0.655 on the same golden set,
+# which is what tests/eval/baseline.json records at the current pins. The
+# second set is what a real user sees, so flooring against the first was
+# measuring the corpus rather than the retrieval — three of the four old floors
+# fail outright once the corpus is realistic. Each floor now sits ~10% below
+# its measured value: loose enough that a query or two cannot flake the build,
+# tight enough that a collapse fails it. Regression-vs-baseline stays relaxed
+# (--regression-eps 1); the absolute floors are the real gate. (It was relaxed
+# while the corpus tracked upstream HEAD. tests/eval/taps.txt pins a commit per
+# row now, so that is no longer the reason.)
+#
+# ci.yml and eval-corpus-refresh.yml run this same eval_retrieval.py call.
+# tests/unit/test_eval_corpus.py parses all three with the script's own parser
+# and fails the build on any difference, -k and --golden included.
 eval:
 	PYTHON=$(PY) BOOST_HOME=$(EVAL_HOME) bash scripts/ensure_eval_corpus.sh
 	BOOST_HOME=$(EVAL_HOME) $(PY) scripts/eval_retrieval.py --build -k 10 \

{boost_skill_cli-1.2.131/boost_skill_cli.egg-info → boost_skill_cli-1.2.133}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boost-skill-cli
-Version: 1.2.131
+Version: 1.2.133
 Summary: boost — Homebrew for AI coding skills
 Author: Jonathan Reyes
 License-Expression: Apache-2.0
@@ -62,9 +62,9 @@ Antigravity CLI (`agy`) is wired up the same way: skills link into
 
 ```bash
 pipx install boost-skill-cli
-boost quickstart              # taps the 7 starter registries, loads prebuilt vectors
-boost search tdd              # keyword + semantic, fused
-boost install tdd-workflow    # → every agent, version-pinned, one lock file
+boost quickstart                       # taps the 7 starter registries, builds the index
+boost search tdd                       # ranked by full-content BM25
+boost install test-driven-development  # → every agent, version-pinned, one lock file
 ```
 
 (No `pipx` yet? macOS ships neither it nor a new enough Python — run the
@@ -320,11 +320,18 @@ If you tapped heavily before this landed, `boost compact` narrows the clones you
 already have, offline and with no loss of search coverage:
 
 ```bash
-boost compact --dry-run     # what it would reclaim
-boost compact               # narrow every clone in place
-boost compact --reclone     # also drop downloaded git objects (needs network)
+boost compact --dry-run            # what narrowing would reclaim
+boost compact                      # narrow every clone in place
+boost compact --reclone --dry-run  # what a re-clone would remove first
+boost compact --reclone            # also drop downloaded git objects (needs network)
 ```
 
+Narrowing removes only files git tracks. A file you dropped into a clone by
+hand stays where it is, so `--dry-run` does not count it. `--reclone` deletes
+the whole clone, so its preview does count those files, plus the `.git` it
+drops. It gives no net figure, because the size of the fresh clone is up to
+the remote.
+
 boost indexes three item kinds from the same registries, and they install to
 different places. A skill is copied into the canonical store and symlinked out.
 A rule is materialised into each agent's context file (`~/.claude/CLAUDE.md`,

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/README.md

@@ -20,9 +20,9 @@ Antigravity CLI (`agy`) is wired up the same way: skills link into
 
 ```bash
 pipx install boost-skill-cli
-boost quickstart              # taps the 7 starter registries, loads prebuilt vectors
-boost search tdd              # keyword + semantic, fused
-boost install tdd-workflow    # → every agent, version-pinned, one lock file
+boost quickstart                       # taps the 7 starter registries, builds the index
+boost search tdd                       # ranked by full-content BM25
+boost install test-driven-development  # → every agent, version-pinned, one lock file
 ```
 
 (No `pipx` yet? macOS ships neither it nor a new enough Python — run the
@@ -278,11 +278,18 @@ If you tapped heavily before this landed, `boost compact` narrows the clones you
 already have, offline and with no loss of search coverage:
 
 ```bash
-boost compact --dry-run     # what it would reclaim
-boost compact               # narrow every clone in place
-boost compact --reclone     # also drop downloaded git objects (needs network)
+boost compact --dry-run            # what narrowing would reclaim
+boost compact                      # narrow every clone in place
+boost compact --reclone --dry-run  # what a re-clone would remove first
+boost compact --reclone            # also drop downloaded git objects (needs network)
 ```
 
+Narrowing removes only files git tracks. A file you dropped into a clone by
+hand stays where it is, so `--dry-run` does not count it. `--reclone` deletes
+the whole clone, so its preview does count those files, plus the `.git` it
+drops. It gives no net figure, because the size of the fresh clone is up to
+the remote.
+
 boost indexes three item kinds from the same registries, and they install to
 different places. A skill is copied into the canonical store and symlinked out.
 A rule is materialised into each agent's context file (`~/.claude/CLAUDE.md`,

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/boost_cli/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '1.2.131'
-__version_tuple__ = version_tuple = (1, 2, 131)
+__version__ = version = '1.2.133'
+__version_tuple__ = version_tuple = (1, 2, 133)
 
 __commit_id__ = commit_id = None

{boost_skill_cli-1.2.131 → boost_skill_cli-1.2.133}/boost_cli/cli.py

@@ -118,7 +118,7 @@ COMMANDS = [
     ("health",      "chk", "quality", "Dashboard of skill-environment health"),
     ("trust",       "chk", "quality", "Manage signing keys & verify tap provenance"),
     # Configuration (13)
-    ("quickstart",  "cfg", "quickstart", "Tap the starter registries and load prebuilt vectors"),
+    ("quickstart",  "cfg", "quickstart", "Tap the starter registries and index them for search"),
     ("config",      "cfg", "configuration", "Display or modify boost configuration"),
     ("clean",       "cfg", "configuration", "Clear stale caches & broken symlinks"),
     ("compact",     "cfg", "configuration", "Shrink tap clones to the files boost indexes"),
@@ -199,6 +199,7 @@ def print_help() -> None:
     print(out.c(
         "Options:  -V/--version   -v/--verbose   --debug   -q/--quiet",
         out.DIM))
+    _print_first_run_line(cols)
 
     width = max(len(n) for n, _, _, _ in COMMANDS)
     for idx, (gkey, (_icon, title, desc)) in enumerate(GROUPS.items()):
@@ -223,6 +224,35 @@ def print_help() -> None:
     print(out.c("%d commands · %d groups" % (len(COMMANDS), len(GROUPS)), out.DIM))
 
 
+def _print_first_run_line(cols: int) -> None:
+    """Name where setup lives, but only on a machine that has none.
+
+    This screen read no state at all, so it was byte-for-byte identical on a
+    virgin machine and a working one, and a newcomer met 81 commands with
+    `quickstart` sitting at line 80, in the same dim body type as the other 80.
+    The pointer is one line, it appears only while `config.first_run()` holds,
+    and it disappears the moment there is a tap: a permanent banner would be
+    noise for every user past their first minute, which is the reason there
+    wasn't one.
+
+    Asked of `config`, which help has already loaded for its log level, so the
+    question costs `boost --help` no module it was not importing anyway. Any
+    failure is swallowed — a broken config must cost the user a hint, never
+    their help screen.
+    """
+    with contextlib.suppress(Exception):
+        from .core import config
+        if not config.first_run():
+            return
+        print()
+        # Wrap, then colour: `out.role` brackets its argument with a start code
+        # and a reset, so colouring first and splitting after leaves line one
+        # unterminated. The backtick span stays one atomic token, because a
+        # `boost quickstart` folded across two lines is not copy-pasteable.
+        for line in out.wrap(config.FIRST_RUN_HINT, max(cols, 20)):
+            print(out.role(line, "muted"))
+
+
 def print_command_help(name: str) -> int:
     # `main`'s own aliases (-h/--help, -V/--version/version, help itself)
     # never appear in COMMANDS, so `boost help --help` used to fall straight