Skip to content

Concise skills rollout: plugin-quality (3 skills, 2 model-invoked) #6588

Open

Description

@kyle-sexton

Part of #6476. Rewrite plugin-quality's skills to the skill criteria in one draft PR, following the per-plugin runbook (/playbooks:skill-authoring, reference/rewrite-a-plugin.md). The runbook and the rewrite protocol it cites decide what ships; this issue holds only what is specific to the plugin.

The plugin today

As of origin/main on 2026-10-24:

  • Skills: audit, report (model-invoked); setup (user-invoked).
  • Outcome eval case directories under the plugin: 0. Probe files under probes/: 0.
  • 1 agent definition(s): out of scope for this skills pass unless the umbrella's agent pilot has landed.

Done when

  • /skill-quality:check plugins/plugin-quality on the PR head reports no FAIL that the run on main does not.
  • /evals:validate passes on plugins/plugin-quality's eval suite at the PR head.
  • Each model-invoked skill has a probe file: plugins/plugin-quality/probes/audit.json, plugins/plugin-quality/probes/report.json.
  • The PR body holds a per-skill decision table, description and body decided separately, with the before and after numbers behind each row.
  • The PR body names every skill or case left unmeasured, with the reason.
  • The plugin's version or changelog entry is in the PR, per the plugin's release mode.

Eval spend and review

  • Eval passes within prepaid API credit are pre-approved (owner decision Q46, 2026-10-24), at the --max-cost-usd 5 per-pass ceiling set on #6476 on 2026-10-18.
  • No person gates the PR: the owner's 2026-10-24 decision (C12) leaves only an intent mismatch during the build or a security issue at review as stops.

Generated by Claude Code

Activity

  1. added
    needs-triageNot yet classified. Floor until a type and one priority tier are set.
    on Oct 7, 2026
  2. github-actions commented on Oct 7, 2026

    @github-actions
    Contributor

    Proposed work class: work-class: scoped
    Basis: automated intake triage; the lanes App applied this class.

    Size: Medium
    Complexity: Medium
    Uncertainty: Known
    Flags: none
    Blocked by: none

    I understand this as the per-plugin skills rewrite of plugin-quality (audit, report, setup), done in one draft PR under #6476. It follows the rewrite-a-plugin runbook. The PR needs probe files for the two model-invoked skills, a per-skill decision table, a list of anything left unmeasured, and a version or changelog entry. Eval spend is on prepaid credit, so I've added no cost flag. The done-when items are specific enough to act on, and I found nothing missing and no likely duplicates.

    Automated triage, applied by the lanes App.

  3. added
    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.
    on Oct 7, 2026
  4. removed
    needs-triageNot yet classified. Floor until a type and one priority tier are set.
    on Oct 12, 2026
  5. added theissue type on Oct 12, 2026
  6. removed
    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.
    on Oct 24, 2026
  7. kyle-sexton commented on Oct 24, 2026

    @kyle-sexton
    MemberAuthor

    Released by the owner's HITL session (2026-10-24, C12): it waited only on a removed gate. Intake triage re-runs on it to set its fields, then admission takes it.

  8. github-actions commented on Oct 25, 2026

    @github-actions
    Contributor

    The triage sweep dispatched intake triage for this issue (attempt 1 of 2).

  9. added
    status: readyTriaged, unblocked, and fully specified; eligible to pick up.
    work-class: scopedA briefed fix or small feature; blast radius bounded by the brief, tests exist.
    admission-deferredReady for the issue lane; work-unblock admits it when queued work is under the queued cap
    agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.
    and removed
    admission-deferredReady for the issue lane; work-unblock admits it when queued work is under the queued cap
    on Oct 25, 2026
  10. melodic-automation-lanes commented on Oct 25, 2026

    @melodic-automation-lanes
    Contributor

    The lanes App's pull request #9999 merged with Refs:, so this issue stays open and nothing starts the lane on it again. work-unblock removed status: ready and agent-ready, so the hourly triage sweep sends it back to intake triage, which decides what work is left and admits it again.

  11. added
    status: readyTriaged, unblocked, and fully specified; eligible to pick up.
    admission-deferredReady for the issue lane; work-unblock admits it when queued work is under the queued cap
    agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.
    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.
    and removed
    status: readyTriaged, unblocked, and fully specified; eligible to pick up.
    agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.
    admission-deferredReady for the issue lane; work-unblock admits it when queued work is under the queued cap
    on Oct 25, 2026
  12. melodic-automation-lanes commented on Oct 25, 2026

    @melodic-automation-lanes
    Contributor

    The work-resolve-issue lane stopped partway through this issue and needs a person.

    Why, in the run's own words:

    This lane cannot run the measurements the rewrite protocol needs before anything ships. The runbook (plugins/playbooks/skills/skill-authoring/reference/rewrite-a-plugin.md) and skill-criteria.md 'Measurement' step 6 ('ship only parts that were measured'; 'A cut to an existing skill ships only when its evals show no loss') require several runs: /evals:plugin-eval preflight, a smoke pass on the before version, /evals:validate, judge calibration, before and after eval passes on each target model, /skill-quality:check measure-invocation over a held-out probe set written outside the repository, and /docs-hygiene:compress compare with two labelers. The issue's Done-when also needs /skill-quality:check and /evals:validate results and a decision table backed by before and after numbers. The CI lane loads only the work-items, discovery and knowledge plugins, and its Bash runs only the named wrappers, so the evals, skill-quality and docs-hygiene skills and the claude plugin eval CLI are not available here. An unmeasured rewrite cannot ship under the protocol. Writing cases alone, without the smoke pass and freeze, would leave half-finished work that a later run would have to redo, so nothing was started. This issue needs a session that has the evals, skill-quality and docs-hygiene plugins and the prepaid eval credit (Q46, --max-cost-usd 5 per pass).

    What the run says it did not do:

    • Inventory and before-state /skill-quality:check plugins/plugin-quality run: the skill-quality plugin is not loaded in this lane
    • Outcome cases via /evals:design, /evals:validate, judge calibration, and the before-version smoke pass with freeze: the evals plugin and the paid eval runner are not available in this lane
    • Held-out probe set and /skill-quality:check measure-invocation triggering runs: not available in this lane
    • Rewriting the audit, report and setup skills: not started, because the protocol ships only measured parts and nothing could be measured here
    • Before and after measurement, the /docs-hygiene:compress compare meaning diff, and the per-skill decision table for the PR body: not available in this lane
    • plugin-quality changelog fragment: not written, since nothing in the plugin changed

    Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/40208741583

    To retry, remove needs-human and add agent-ready again.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.priority: mediumReal value, no hard deadline; normal backlog flow.status: readyTriaged, unblocked, and fully specified; eligible to pick up.work-class: scopedA briefed fix or small feature; blast radius bounded by the brief, tests exist.

    Type

    Fields

    No fields configured for Task.

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions