Repository navigation
Closed
17 / 1717 of 17 issues completedClosed
17 / 1717 of 17 issues completed
Copy link
Labels
spec-containerSpec container: sub-issues are its work items (/work-items:decompose and /work-items:ship).Spec container: sub-issues are its work items (/work-items:decompose and /work-items:ship).
Description
Follow-ups carried out of the code-metrics plugin's first release (PR #3766), none of which blocks that PR. Each is independent; take them as separate PRs.
Operator decisions
- Q24, the two convention adopter rows. The marketplace's per-plugin conventions registry needs the plugin's adopter rows once the operator settles which conventions it adopts; the README's "Known gaps" section says so until then.
- Q23, a
skill-doctorrow indocs/native-surfaces/records.json. Needs anupstream-sourcepin (theanthropics/claude-codeCHANGELOG commit SHA for the 2.1.261/skill-doctoraddition), which was outside the build session's repository scope. Theaudit-skill-visibilityrouting already names the surface behind a presence gate (claude-ops 0.42.8).
From the plugin-quality audit (packet run 20260905T173832Z)
- Generate
reference/config.md's key table fromscripts/config-defaults.jsonso the two surfaces cannot drift (the setup template is already bound to the defaults by a test; the reference and the SKILL.md bodies remain prose copies). - Root-relative
filefields in thecode-metrics/v1schema. Today every path is relative to the working directory the audit ran from; a consumer comparing two reports from different directories has to normalize. cobertura.pyreads only the first<source>root; gcovr emits several for multi-root builds. Read all of them (theaudit-coverageGotchas document the limit).- The coverage join's trailing-name fallback can bind
A.runtoB.runinside one file when the artifact carries only the short name; prefer a range-overlap tie-break. - Invoke each skill through the Skill tool in a consumer repository to observe pre-computed context rendering, the quoted-path
allowed-toolsmatch, and description triggering; the build session only ran the scripts directly.
From the phase verifiers
- After a total sanctioned-replication exclusion,
summary.filescounts surviving rows only while the scope header shows the file count; consider counting the excluded instances' files in the summary. audit-type-debt.test.shhardcodes the collector-name list for its tool-free PATH; a ladder tool added later must be added there too.- Three skill descriptions sit within 20 codepoints of the 1024-codepoint spec field cap (
principles,audit-complexity,audit-coverage).
Out of scope by design (recorded, not follow-ups)
- The C# complexity lane (deferred behind a runtime probe of
Microsoft.CodeAnalysis.Metrics). - A Bash function-range collector for CRAP.
Metadata
Metadata
Assignees
Labels
spec-containerSpec container: sub-issues are its work items (/work-items:decompose and /work-items:ship).Spec container: sub-issues are its work items (/work-items:decompose and /work-items:ship).
Type
Fields
No fields configured for issues without a type.
Activity
feat(code-metrics): add the code-metrics plugin, seven read-only meas…
fix(code-metrics): stop an out-of-scope package crediting a scoped fi…
kyle-sexton commentedon Sep 6, 2026
This was generated by AI during triage.
Claimed for triage by autonomous lane instance
vm-triage-202609050304. Lease expires 2026-09-06T19:40:00Z. This is a triage claim only, not an execution claim; the item stays unassigned and available to the worker lane once briefed.Generated by Claude Code
kyle-sexton commentedon Sep 6, 2026
This was generated by AI during triage.
Triage verdict: decomposed
Category: Task. State: oversized for a single brief → decomposed into ten native sub-issues. Parent role:
work-mapcontainer; it tracks the set and is not itself claimable work.Routed to decomposition rather than briefed as one item because the body carries ten independent follow-ups spanning parsers, a published schema, documentation binding, test hygiene, an operator decision, and a manual QA pass. The issue says so itself: each is independent and wants its own PR. A single brief would have been unworkable, and a cold agent picking up the parent could not have known where to stop.
Verification before slicing
Spot-checked the claims against current
mainrather than taking them on trust. Confirmed: the plugin ships the seven skills described; the configuration defaults file and the configuration reference both exist as separate surfaces, the reference table hand-maintained; the known-gaps section records the deferred convention adopter rows; the native-surfaces record carries the note recording the owed upstream pin and the reason it could not be taken.The Cobertura defect was confirmed in the code, not just in the report: the parser collects every declared source root into a list and then applies only its first element to relative class filenames. The remaining items were read as described and matched their surfaces.
The ten slices
Frontier first, unblocked items at the top.
scopedscopedscopedscopedscopedmechanicalscopedstructuralscopedread-onlySeven are autonomous-eligible and unblocked; three need a human.
Why the three are human-gated
Sequencing note
#3842 touches code adjacent to #3839 and #3840; taking it after them avoids a needless conflict. #3848 is most informative once #3845 and #3841 have landed, since both change what a QA pass would observe. Neither is a hard dependency, so no blocking edges were set.
Not filed
The two items the parent records as out of scope by design (the deferred C# complexity lane and the Bash function-range collector for the composite metric) were deliberately not turned into sub-issues. The parent records them as decisions, not as work, and filing them would convert a recorded decision into a backlog obligation nobody agreed to.
This parent stays open as the container. Per the decomposition contract it was not closed and its body was not modified.
Triage claim released by
vm-triage-202609050304.Generated by Claude Code
12 remaining items
fix(code-metrics): ground the principles skill in its primaries and a…
fix(code-metrics): five audit-coverage follow-ups from the 0.1.0 build (
docs(out-of-scope): keep issue numbers off line starts (#3768)
feat(code-metrics): files_excluded_only and skill-prose default pins (#…
kyle-sexton commentedon Sep 29, 2026
From the 2026-09-29 audit of the Cursor run. This work-map stays open. Recording what PR #5142 did and did not satisfy, so lanes do not re-pick finished bullets.
scripts/check-code-metrics-skill-prose.py(14 phrases, not the 9 the PR body states). The PR body's "closes two README known-gaps" is inaccurate: its diff replaces one gap.summary.files_excluded_onlywith no open issue; it is additive withincode-metrics/v1. The audit fix in fix(code-metrics): partial-label the type-debt lane row and dedupe the excluded-only count #5308 computes that count once in render.Lanes: check sub-issue state before picking up a parent bullet.
fix(code-metrics): partial-label the type-debt lane row and dedupe th…
docs(conventions): declare code-metrics divergences in its convention…
kyle-sexton commentedon Oct 1, 2026
Closing at the owner's direction on 2026-10-01 ("If there's anything that's old, not relevant, or not needed, we can close those"), after a read-only check of every child.
All ten follow-up slices (#3839 to #3848) and the later audit slices (#3993 to #3998, #4067) are closed with merged PRs. The config-reference item (#3841) shipped as a verify gate, not a generator: #3912 added scripts/check-code-metrics-config-reference.py and wired it into ci.yml, pinning the key set and defaults against config-defaults.json; the prose column stays hand-written on purpose. The two out-of-scope items (C# complexity lane, Bash function-range collector) were never follow-ups.