Review of the ADR-0020 gate found four ways it could exit 0 without measuring, and
one way it hard-failed a repo it had no business failing. On a gate shipping hot
with no baseline, a silent pass is the worst outcome available and a false block is
the second worst.
Consumer resolution was the blocker. _authoring_root() fell back to the nearest
.git, so it returned truthy in ANY git repo; _collect_authoring_root() then
contributed nothing and the deployed-tree branch was dead code in precisely the
consumer case it exists for. A consumer repo routing to an installed sibling got an
unblockable ERROR, and deleting .git "fixed" it. It now keys on which of the two
walk-up passes matched. A name-count delta was tried first and is wrong: a
single-plugin monorepo re-collects its own package and adds no new name, so the
delta reads zero and drags the deployed trees — including a global ~/.claude — back
into the universe. That reintroduces the install-dependence ADR-0020 forbids, one
layer down.
The three silent passes: an indented `---` inside a block scalar truncated the
frontmatter and reclassified the rest of the description as body; a non-string
description was str()-coerced, so `description: true` measured as the four-character
"True"; and an unterminated fence blanked the rest of the body, disabling the
ERROR-tier references/ check and the gotcha counts.
Two measurement defects came with them. The awk line/word counts discarded awk's
exit status, so an unreadable file passed both spec ceilings in total silence, and
awk NR/NF disagreed with the audit script's splitlines()/split() on Unicode
whitespace — the "fix one gate, get blocked by the other" bug, on the two axes the
differential test deliberately excluded. Both counts now run in the Python block
that already reads the file. A type error also no longer reports itself as a syntax
error.
Also: glob metacharacters in the checkout path silently disabled the resolver;
re.I was applied to some extraction patterns and not others; agent-audit missed
`tools:` written as a YAML block sequence, the shape Copilot files use; and a
nonexistent agent file raised a bare FileNotFoundError instead of a diagnostic.
The shared resolver block stays byte-identical across all three scripts. Corpus
output is unchanged — 26 description FAIL, 9 body FAIL, 2 dangling, 0 missing
references, 58 SUGGESTIONs — so no documented count moves.
Refs: #99
ADR: 0020
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015W3iwF9ncfRZddGBxsMCYi
Three ways the gates could report green having measured nothing. All three were
invisible to a passing test suite, because pre-commit prints nothing at all for a
hook that exits 0 — a gate that declines to check and a gate that checked and
passed produce the identical signal.
- A UTF-8 BOM, a leading blank line, a trailing space after a `---` marker or
CRLF line endings defeated the `^---\n` frontmatter matcher. Every ADR-0020
check was then skipped and the file passed: measured at the time, a
550-character description with a 1,000-word body exited 0 behind a BOM.
All four shapes are now tolerated, and frontmatter that genuinely cannot be
parsed is a hard ERROR rather than a silent skip.
- An agent file with a valueless `description:` followed by another key let a
line regex capture the *next* key, which looked non-empty, so the
missing-or-empty branch never fired and every gate below it early-returned on
the empty folded value — zero output, exit 0, on a blocking gate. The one
field this contract is entirely about was the one field a gate could fail to
notice was absent. Presence is now decided on the YAML-folded value and
nowhere else, and a missing or empty description is a hard FAIL in all three
validators.
- The hand-rolled frontmatter fallback disagreed with PyYAML across the FAIL
boundary on folded scalars, so which reader happened to be available decided
the verdict. A fallback that mis-parses a scalar shape reports a vacuous pass,
which is worse than not running, so it is deleted: python3 and PyYAML are hard
requirements that fail loudly with an install pointer.
Boundary-target resolution no longer derives its universe from its own location.
A `${BASH_SOURCE}`-relative repo root leaked this repo's 39-skill universe into
every consumer repo running the hook through pre-commit, so a consumer skill
routing to `skill-audit` resolved against a plugin it had never installed. The
interim form resolved through `.claude/` and `.agents/`, which are gitignored
`apm install` output — the same commit reported 2 dangling targets on a machine
that had run the install and 6 on a fresh clone. Resolution now walks up from the
file being checked to an authoring root (nearest ancestor holding
`plugins/*/.apm/{skills,agents}`, else the nearest `.git`, in two passes so a
nested `.git` cannot outrank a real monorepo root); the universe is every skill
and agent under `<root>/plugins/*/` plus the file's own apm package and that
package's declared `dependencies.apm`. Deployed trees are consulted only when no
authoring root exists at all — the consumer case. One commit now gets one verdict,
which a gate shipping hot with no baseline file has to.
Narrowed in the same pass: a routing target inferred from the prose boundary form
and corroborated by nothing else reports at SUGGESTION instead of blocking. A
blocking check with no escape hatch is the wrong trade when the inference from
prose is the weak part of it.
New deterministic checks, all previously untested or absent: every
`references/<file>.md` a body names must exist (ERROR — a broken pointer is not a
style opinion); a description with no boundary clause at all, a Gotchas section
over five entries, and a Gotchas section over 25% of the body are SUGGESTIONs.
Where no universe can be determined the target check prints `INFO ... DID NOT
RUN` rather than passing quietly. Each prose-scanning check needed its own
false-positive fix — a fenced example of a Gotchas section was being read as the
section itself — and those fixes are pinned rather than assumed.
The resolver is one block copied verbatim into all three scripts between
BEGIN/END markers, because a cache-installed plugin's scripts cannot read outside
their own plugin directory. Nothing asserted the copies were still identical; a
one-line edit to a single copy passed every constant-agreement assertion, since
constants are not what drifts.
Tests land here rather than in a later commit. The existing suites assert the old
behaviour and go red against these scripts, so splitting them would leave a commit
whose own `run-tests` pre-push gate fails in isolation.
Refs: ADR-0020
Skill name+description pairs are preloaded into every session, costing
~6,200 tokens across 39 skills before any skill is invoked. The authoring
rules mandated that growth: skill-author:104 and description-quality.md:21
both required padding, while skill-author:102 (the deflating rule) had no
FAIL condition behind it.
Gates (blocking, no baseline file):
- description 250 chars SUGGESTION / 400 FAIL, measured on the folded
YAML value
- body-only 600 words SUGGESTION / 900 FAIL, independent of the unchanged
whole-file 2770-word / 500-line spec backstop
- every boundary-clause routing target must resolve to a real skill or
agent; catches skill-improve, neuledge-context and gitea-labels
- agents take the description gates but deliberately no body gate; a test
pins that absence
Vale: DescriptionOpener widened to ^This\b, new CompositionNote rule
banning architecture notes from descriptions. 10 hits, 0 false positives.
Kyberforge's own four skills retrofitted: descriptions 3,364 -> 938 chars
(-72%), bodies 8,306 -> 2,487 words (-70%), all via the apm-workflow
dispatch pattern. Fixes the skill-improve dangling route and the
agent-author misroute to manual review.
Also fixes a pre-existing false positive where any line-initial 'read '
was flagged as interactive input, which had already caused two scripts to
be rewritten around it.
Refs: ADR-0020
MAX_WORDS=2900 was calibrated to the corpus median density and carried no
margin: at the densest observed 7.22 chars/word (~1.81 tokens/word) it permits
~5,240 tokens against the 5,000 it proxies for. 2770 holds the worst observed
density under the ceiling. The largest SKILL.md is 2,489 words, so the change
costs nothing today — 281 words of margin — and the header comment now argues
the new calibration rather than swapping the digits.
Both enforcement points move together, and a new test asserts they agree, since
a SKILL.md passing its own audit while the commit hook blocks it is the
disagreement this pair exists to prevent.
CONTEXT.md is deliberately left ungated: it is 2,816 words, and gating it would
block the build. Recorded here so the omission reads as a decision rather than
an oversight.
skill-audit's manual-fallback path listed only the line ceiling, so an agent
taking that path passed an oversized SKILL.md the hook then rejected. The word
ceiling is now named alongside it. agent-audit is deliberately unchanged: the
size hook scopes to SKILL.md only and agent-audit's validate.sh has no word
gate, so claiming it there would be false.
The Vale research doc still showed the MDX {/* vale off */} form under a
Markdown heading, contradicting CONTEXT.md and vale-run's troubleshooting
reference — that form suppresses nothing in plain .md. Fixed in both places it
appeared.
tests/run-tests.sh used mapfile (bash 4.0+) with unguarded array expansion,
though AGENTS.md tells contributors to run it and macOS ships bash 3.2. It now
collects via a while-read loop over process substitution and guards every
expansion. The newline-delimited find|sort pipeline is kept rather than -print0
with sort -z, whose BSD portability is the weaker link, and which matches
mapfile -t's previous behaviour exactly.
Refs: #85
ADR: 0013
Three divergences between what the audit skills claim and what the hooks
enforce, each of which fails silently rather than loudly:
- `skill-size-check.sh` blocked at 2900 words while `validate.sh` checked only
the 500-line ceiling, so `/skill-audit` could report a skill ready to ship
that the commit hook then rejected. `validate.sh` now checks the same pair on
the same inclusive terms; the constants are duplicated with a comment naming
the other file, because a plugin skill's scripts cannot read outside the
plugin directory once installed to the cache.
- Both audit skills' Step 1 passed `--config assets/vale/.vale.ini`, which is
redundant (the wrapper self-locates its sibling config) and fragile: an agent
that resolves the script path against the skill directory but not the config
path gets E100, exit 2, which the surrounding fallback clause misreads as
"vale unavailable" and downgrades to full LLM judgment with no signal.
- The external-consumer test registered only the two Vale hooks, never the
third shipped hook, so a lost executable bit would have broken every consumer
while the local suite stayed green. Verified by mutation: `chmod 644` on the
copied script now turns three passes into two failures.
Also corrects the size hook's calibration comment, which claimed ~5.7-6.5
characters per word against a corpus whose measured median is 6.79 — the stated
upper bound sat below the median, so the "calibrated with margin" claim was
inverted for prose-dense files. MAX_WORDS is unchanged pending a decision; the
comment is now explicit that the gate holds under 5,000 tokens for typical
prose density, not for any file.
Refs: #85
- validate.sh: move FAIL lines and failure summary to stdout; stderr
reserved for fatal script errors only (missing SKILL.md, bad args)
- skill-audit SKILL.md: replace concrete plugins/kyberforge/skills/...
example with abstract placeholder to fix meta-circularity
- skill-write SKILL.md: rephrase placeholder section-heading instruction
to remove embedded FILL IN: from a code span, clearing validator false positive
- skill-improve SKILL.md: wrap Step 2 root-cause example in a text fence
- skill-write assets/templates/README.md: update Files table to individual-
file rows so skill-audit can verify per-file coverage
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Move validate.sh ownership to skill-audit/scripts/ — it is the canonical
structural validator; skill-write now delegates Step 5 to /skill-audit
- Add skill-write/references/scripts.md and deployment-modes.md for progressive
disclosure of package runner patterns and plugin cache isolation rules
- Fix skill-audit Step 1 cross-skill path reference (was repo-absolute, now
skill-relative); add manual fallback for sandboxed/Bash-denied contexts
- Scope Step 2 "read every file" to exclude binaries and unreferenced files
- Fix new-skill.sh next-steps output to reference /skill-audit instead of
the removed validate.sh
- Remove stale Dependencies section from skill-audit README; flip dependency
arrow — skill-write depends on skill-audit, not vice versa
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
