ci(skillsaw): add lint job, Makefile targets, and repo config - #754
Conversation
PR Summary by QodoCI: run skillsaw lint in strict mode and add local Make targets
AI Description
Diagram
High-Level Assessment
Files changed (3)
|
|
🤖 Finished Review · ✅ Success · Started 5:01 PM UTC · Completed 5:17 PM UTC Commit: |
Code Review by Qodo
1.
|
![qodo-code-review[bot]](https://avatars.githubusercontent.com/in/484649?s=60&v=4){kind=small}
ReviewFindingsMedium
Low
Next steps:
Previous runReviewFindingsMedium
Low
Next steps:
Previous run (2)ReviewFindingsMedium
Low
Previous run (3)ReviewFindingsMedium
Low
Previous run (4)ReviewFindingsMedium
Low
Previous run (5)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (6)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (7)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (8)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (9)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (10)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (11)ReviewFindingsMedium
Low
Previous run (12)ReviewFindingsMedium
Low
Previous run (13)ReviewFindingsMedium
Low
Previous run (14)ReviewFindingsMedium
Low
Previous run (15)ReviewFindingsMedium
Low
Previous run (16)ReviewFindingsMedium
Low
Previous run (17)ReviewFindingsMedium
Low
Previous run (18)ReviewFindingsMedium
Low
Previous run (19)ReviewFindingsHigh
Medium
Low
Next steps:
Previous run (20)ReviewFindingsMedium
Low
Previous run (21)ReviewFindingsMedium
Low
|
![fullsend-ai-review[bot]](https://avatars.githubusercontent.com/in/3478501?s=60&v=4){kind=small}
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
fullsend-ai-review
Bot
added
the
requires-manual-review
waynesun09
left a comment
waynesun09
left a comment
There was a problem hiding this comment.
Automated review sweep: 4 findings (2 HIGH, 2 MEDIUM) posted as inline comments.
|
These three all landed as separate inline threads — handled them there (protected-path is intentional, indentation fixed, strict-mode comment reworded). |
|
Went through these: Doc gaps — added Scope vs #188 — I see the issue only asked for evaluation, but I'd count the PR itself as the evaluation: building the strict lint job out and watching it catch real issues in review (the ones fixed inline on this PR) is the signal that it finds meaningful problems and is worth running ongoing. Comfortable moving straight to CI integration rather than splitting that into a separate PR. persist-credentials — answered on the dedicated thread for that one. PR title — keeping |
|
🤖 Finished Review · ✅ Success · Started 1:16 PM UTC · Completed 1:32 PM UTC Commit: |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
waynesun09
left a comment
waynesun09
left a comment
There was a problem hiding this comment.
Automated review sweep: 2 findings (1 HIGH, 1 MEDIUM) posted as inline comments.
b703e8c to
c2ab7a5
Compare
|
🤖 Finished Review · ✅ Success · Started 9:39 PM UTC · Completed 9:54 PM UTC Commit: |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
fullsend-ai-review
Bot
removed
the
requires-manual-review
|
🤖 Finished Fix · ✅ Success · Started 9:55 PM UTC · Completed 10:01 PM UTC Commit: |
🔧 Fix agent — iteration 1 (bot-triggered)Addressed 2 of 3 review findings. Fixed invalid uvx syntax in Makefile (== to @) and added lint.yml to README Workflows table. Disagreed with 1 finding (persist-credentials consistency) per reviewer's own suggestion to handle in a follow-up PR. Fixed (2):
Disagreed (1):
Tests: passedNext steps:
|
|
🤖 Finished Review · ✅ Success · Started 10:02 PM UTC · Completed 10:14 PM UTC Commit: |
Superseded by updated review
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
fullsend-ai-review
Bot
added
the
requires-manual-review
|
🤖 Finished Review · ✅ Success · Started 11:32 PM UTC · Completed 11:45 PM UTC Commit: |
|
🤖 Finished Review · ✅ Success · Started 12:54 AM UTC · Completed 1:05 AM UTC Commit: |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
waynesun09
left a comment
waynesun09
left a comment
There was a problem hiding this comment.
Review-only sweep on head 8927c03 — 2 findings (1 HIGH, 1 MEDIUM) posted inline. Not approving or requesting changes; leaving disposition to the author.
|
🤖 Finished Review · ✅ Success · Started 2:06 AM UTC · Completed 2:27 AM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $8.08 |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
fullsend-ai-review
Bot
removed
the
requires-manual-review
Adds a skillsaw job to the lint workflow, running the linter in --strict mode on every PR and push to main. Adds `make lint` and `make lint-fix` targets so contributors can run the same checks locally. Adds .skillsaw.yaml, tuned for this repo: - Disables marketplace-json-valid and marketplace-registration. plugins/gopls-lsp is an internal fullsend harness plugin config (see harness/code.yaml), not a Claude Code marketplace plugin, so these rules produced false positives. - Raises the context-budget skill thresholds to warn 16000 / error 20000, above code-implementation and pr-review's current sizes. Both are dense, production-tuned orchestration procedures; --strict turns warnings into failures, so the thresholds are set high enough that neither file triggers a violation at either severity. This builds on the skill content fixes in the preceding PRs in this stack, evaluating skillsaw per #188. Related to #188 Assisted-by: Claude Opus 4.6 <noreply@anthropic.com>
- Pin skillsaw action to the exact commit SHA (v0.18.0) instead of the v0 tag object, matching the repo's pinact convention. - Fix .skillsaw.yaml exclude block indentation to match the rest of the file. - Correct the marketplace-json-valid/marketplace-registration disable comment: it wasn't gopls-lsp/plugin.json causing the failure, it's that this repo has no .claude-plugin/marketplace.json and isn't a distributable plugin marketplace. - Clarify why strict: false in .skillsaw.yaml doesn't matter for CI. - Document make lint / make lint-fix in README and CONTRIBUTING. Assisted-by: Claude Opus 4.6 <noreply@anthropic.com>
…mment v0.18.0 renamed the Claude Code format rules to claude--prefixed canonical IDs; the checked-in config still used the pre-rename legacy names. Not a functional bug (skillsaw resolves the aliases), but regenerate the keys to match what skillsaw init produces today. Also fix the disabled marketplace-json-valid/marketplace-registration comment, which still cited an already-debunked root cause (gopls-lsp misdetection) instead of the verified one: no top-level .claude-plugin/marketplace.json exists in this repo. Assisted-by: Claude Opus 4.6 <noreply@anthropic.com>
…limit Raising context-budget.limits.skill to warn 16000/error 20000 repo-wide was meant to clear pr-review and code-implementation, but it silently raised the ceiling for every skill file (present or future) and, per review, missed that 3 other files (code-review, fix-review, retro-analysis) were already over the default warn threshold too. skillsaw ships a baseline/ratchet mechanism (baseline_mode: ceiling) built exactly for this: snapshot each currently oversized file's token count as its own ceiling, so it only fails again if it grows past its current size, while every other file keeps the strict default (warn 3000 / error 6000). Revert the limit to defaults and commit .skillsaw-baseline.json with the 5 affected files snapshotted. Assisted-by: Claude Opus 4.6 <noreply@anthropic.com>
CI's skillsaw job scans the pull_request merge commit (head merged into base, per actions/checkout's default ref), not the head branch alone. This PR's base (skillsaw/fix-pr-review) has a slightly different pr-review/SKILL.md than what's on this branch, so the baseline snapshotted from a plain head-branch checkout undercounted its tokens (15,051 vs the merge tree's actual 15,058) and tripped the ratchet in CI. Regenerated by baselining against refs/pull/754/merge directly. Assisted-by: Claude Opus 4.6 <noreply@anthropic.com>
Apply skillsaw's automatic fixes for agentskill-name violations: - Rename forge skills to match directory names (github-forge, gitlab-forge, jira-forge) - Rename issue-labels subdirectory skills to match their directories (github, gitlab, jira) - Add markdown link syntax for bare paths in AGENTS.md Update .skillsaw-baseline.json to suppress context-budget warnings for existing oversized skills (code-implementation now at 9,204 tokens). Add .skillsaw-renames.json to track skill renames for reference updates. Fixes all 7 ERROR-level skillsaw violations. CI checks now pass. Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
The issue-labels skill was renamed to forge-specific names (github, gitlab, jira) in commit 4e91eb4, but the agent markdown files still referenced the old generic name "issue-labels" in their frontmatter skills list and in instruction text. Since these forge-specific skills are loaded dynamically via forge.*.skills in harness YAML (not via agent frontmatter), remove the stale "issue-labels" entry from the skills list in agents/triage.md and agents/review.md. Also update instruction text to reference "label recommendation" skill generically rather than by the old "issue-labels" name, since the actual skill name now varies by forge. Remove the incorrect "issue-labels" → "jira" entry from .skillsaw-renames.json that was added by the auto-fix. The issue-labels skills were renamed to their immediate directory names (github, gitlab, jira), not prefixed variants. This fixes functional test failures where agents couldn't invoke the renamed skills. Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
- Fix invalid skill reference: replace "label recommendation" with "github" in agents/review.md and agents/triage.md - Re-add github skill to frontmatter in both agent files - Remove .skillsaw-renames.json (linter scratch state) and add to .gitignore - Regenerate .skillsaw-baseline.json to fix stale token counts and remove false-positive rename warnings This resolves the HIGH severity issues identified in review: - Invalid skill name with space character (violates AGENTS.md spec) - Stale baseline after rebase causing CI failures - Self-contradictory rename manifest causing false positives - Frontmatter inconsistencies where skills were referenced but not declared Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
The skill renames in commit 4e91eb4 broke the intended design where: - skills/issue-labels/{github,gitlab,jira}/ all share the name 'issue-labels' - skills/{github,gitlab,jira}-forge/ use the names 'github', 'gitlab', 'jira' These names allow skills to be referenced consistently regardless of which forge they're running on, which is the intended behavior. Updated .skillsaw.yaml to document the exceptions and regenerated .skillsaw-baseline.json to exempt these skills from the agentskill-name rule. Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
- Add "Use when..." trigger phrasing to skill descriptions for: - skills/github-forge/SKILL.md - skills/gitlab-forge/SKILL.md - skills/jira-forge/SKILL.md - skills/issue-labels/jira/SKILL.md - Condense skills/fix-review/SKILL.md from 3,937 to under 3,000 tokens: - Streamline verbose sections (Tools reminder, Progress markers, Step 2) - Compress implementation steps while preserving essential guidance - Simplify schema documentation in structured output section - Maintain all critical requirements and constraints Fixes skillsaw lint warnings to pass CI in strict mode. Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
- Revert forge skill names to original (github, gitlab, jira) - Fix agent skill references (issue-labels not github) - Fix triage.md frontmatter field ordering (model before tools) - Restore repo-agnostic test/lint discovery in fix-review - Restore local ref warning in fix-review strategy escalation - Regenerate baseline with pinned skillsaw 0.18.0 Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
Make .skillsaw.yaml the authoritative version source. Makefile now extracts version dynamically. CI comment documents the relationship. Assisted-by: Claude Sonnet 4.5 <noreply@anthropic.com>
The output rules section on line 406 incorrectly referenced the `github` skill for label recommendations. The correct skill is `issue-labels`, which is consistent with the agent's frontmatter and the label recommendations paragraph earlier in the file. Addresses review feedback on #754
Update the "Opening a PR" section to reference both `pre-commit run --all-files` (actionlint, gitleaks, shellcheck, pinact, YAML/JSON) and `make lint` (skillsaw), so contributors know to run both checks locally before pushing. Addresses review feedback from @waynesun09 on #754
Signed-off-by: Ralph Bean <rbean@redhat.com>
efeb110 to
700c695
Compare
|
🤖 Finished Review · ✅ Success · Started 4:04 PM UTC · Completed 4:24 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $5.76 |
|
🤖 Finished Retro · ✅ Success · Started 5:11 PM UTC · Completed 5:23 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $7.14 |
Retro: PR #754 — ci(skillsaw): add lint job, Makefile targets, and repo configTimelinePR #754 (by Review qualityThe review bot's strongest contribution was skill reference integrity checking — it correctly identified broken The human reviewer ( Fix agent failuresBoth fix agent runs (33413947850, 33350710429) failed at "Pre-fetch review body" with Existing issues corroborated by this retro
Proposals skipped (target repo not allowed)File manually or update
|
2 participants
Adds a skillsaw job to the lint workflow, running the linter in
--strict mode on every PR and push to main. Adds
make lintandmake lint-fixtargets so contributors can run the same checkslocally. Adds .skillsaw.yaml, tuned for this repo:
plugins/gopls-lsp is an internal fullsend harness plugin config
(see harness/code.yaml), not a Claude Code marketplace plugin, so
these rules produced false positives.
20000, above code-implementation and pr-review's current sizes.
Both are dense, production-tuned orchestration procedures; --strict
turns warnings into failures, so the thresholds are set high enough
that neither file triggers a violation at either severity.
This builds on the skill content fixes in the preceding PRs in this
stack, evaluating skillsaw per #188.
Related to #188
Assisted-by: Claude Opus 4.6 noreply@anthropic.com
Signed-off-by: Ralph Bean rbean@redhat.com
Stack created with GitHub Stacks CLI • Give Feedback 💬