You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Grades all 41 shipped plugins against the extensibility contract v2.1 (four seams, setup-action requirement, convention-resolution ladder, baked-assumption check, MCP-carry, evals) and records the snapshot as a new audit doc docs/extensibility-contract-grading.md, indexed from the README alongside the smoke-tests record.
Findings
36 of 41 need no net-new work — 9 hook plugins, 5 pure-reference plugins, and 22 behavior plugins are compliant as shipped or their outstanding work is already owned by an existing issue.
5 net-new setup-action gaps — bug-report, discovery, planning, code-tidying, claude-ops each expose a config seam but ship no setup/configure action. One retrofit issue filed per plugin, sub-issue-linked under wave-2 map:
bug-report → melodic-software/medley#1428
discovery → melodic-software/medley#1429
planning → melodic-software/medley#1430
code-tidying → melodic-software/medley#1431
claude-ops → melodic-software/medley#1432
0 baked repo assumptions at runtime; eval coverage deferred to melodic-software/medley#1396; hook migration to melodic-software/medley#1391.
Doc-only change (no marketplace.json / catalog-table touch) — conflict-free with sibling PRs.
Refs melodic-software/medley#1388
Note
Low Risk
Documentation-only; no plugin code, marketplace catalog, or runtime behavior changes.
Overview
Adds docs/extensibility-contract-grading.md, a dated audit snapshot that grades all 41 marketplace plugins against extensibility contract v2.1 (class, config seams, setup actions, convention ladder, baked assumptions, MCP-carry, evals).
The doc concludes 34 plugins need no net-new work, flags 7 behavior plugins with missing or partial setup for their config seams (bug-report, claude-ops, code-tidying, discovery, planning, songwriting, work-items), records 0 baked repo assumptions, and links seven new retrofit issues (#1428–#1435) under wave-2 #1369. It also notes reclassifying songwriting from pure-reference to behavior and defers eval backfill to #1396.
README indexes the new grading doc next to the existing smoke-tests contract doc.
Reviewed by Cursor Bugbot for commit 1981c5c. Bugbot is set up for automated code reviews on this repo. Configure here.
… contract v2.1
Point-in-time grade of all 41 shipped plugins against the four seams, setup-action
requirement, convention-resolution ladder, baked-assumption check, MCP-carry, and evals.
36 compliant/owned; 5 net-new setup-action gaps emitted as retrofit issues.
Refs melodic-software/medley#1388
Review docs/extensibility-contract-grading.md for accuracy and completeness
Review README.md changes
Check alignment with CLAUDE.md guidelines
Post final review findings
Doc-only change grading 41 plugins against extensibility contract v2.1. Counts and cross-references are internally consistent. No security concerns. A few minor editorial issues noted below.
Correctness check
All numerical claims verified against .claude-plugin/marketplace.json (41 plugins) and the grading table:
The implementation row uses plain missing while the 5 net-new gaps use missing (bold). This is a useful signal — bold = net-new issue filed, plain = already owned — but it's never documented in the "Grading dimensions" key. A reader scanning the table could easily treat implementation as compliant. Worth adding a note in the Setup dimension or a table footnote: **missing** = net-new gap (issue filed this session); missing = existing tracked gap.
The intro paragraph defines MCP-carry verdicts as SHIP 0 / STAY 14 / DROP 0. The discovery row reads perplexity/ref/microsoft-learn declared, not shipped (rule 3) which doesn't map to any of those three verdicts. All other MCP-carry cells use the taxonomy or a dash. A label like n/a — declared not shipped or DROP (rule 3, never shipped) would be consistent with the rest of the table.
Links like ["Extensibility contract v2.1 — the four seams"](MIGRATION-PLAYBOOK.md) point to the file root, not the specific section. GitHub resolves #heading-slug fragments across relative links; e.g. MIGRATION-PLAYBOOK.md#extensibility-contract-v21--the-four-seams would land readers at the correct section. Minor quality issue — clicking currently drops the reader at the top of a long playbook.
## Behavior plugins (27) combined with "22 behavior plugins are compliant" in the Verdict summary is accurate but could mislead a quick reader into thinking 5 are uncounted. The existing wording is logically correct once implementation's existing-owned gap is factored in; no change required, but consider (27 total) or a brief parenthetical if readers flag confusion.
Summary
The audit document is accurate, well-structured, and follows the CLAUDE.md "audit snapshot, not durable policy" principle correctly. The date-stamped, Tier-0 fact sourcing, "empirical claims decay" disclaimer, and clear issue tracking are all good practice. The README addition correctly places this alongside the smoke-tests record. The five net-new setup-action gaps and their issues are clearly attributed. Ready to merge; findings above are editorial with no correctness impact.
The reason will be displayed to describe this comment to others. Learn more.
Stale comment
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
Its setup skill persists the catalog overlay but defers the report_dir userConfig to
/plugin configure; knowledge and skill-quality setups interview+persist theirs. Route the
gap to #1419. Addresses Codex PR-review finding.
Refs melodic-software/medley#1388
Review docs/extensibility-contract-grading.md for accuracy and completeness
Review README.md changes
Check alignment with CLAUDE.md guidelines and prior review comments
Investigate Codex finding on machine-health
Post final review findings
Doc-only change; the Codex finding has been addressed. Three minor editorial issues remain open from the prior Claude review plus one new one introduced by the correction commit. No correctness or security concerns.
Codex finding — addressed ✓
Codex flagged that machine-health was incorrectly marked present / compliant in commit 2e5fcc4. The correction commit 657ddf5 fixes this: the row now reads partial with the verdict "setup ships (catalog overlay) but does not interview/persist report_dir — completeness gap + behavior fixes owned by #1419." The Verdict Summary likewise calls machine-health "the lone setup-completeness outlier." Finding resolved.
The Setup action dimension ends with: Present / missing / n/a (no config seam → not required). The correction commit added a fourth grade, partial, for machine-health (line 100), but the key was not updated to document it. Combined with the still-undocumented bold **missing** (net-new gap) vs. plain missing (existing tracked gap) distinction from the prior review, the key now lists three grades while the table uses five distinct values. A reader relying solely on the key will be confused.
Suggested update to line 26:
Present / missing (net-new gap, issue filed this session) / missing (existing tracked gap, issue already owns it) / partial (setup ships but does not cover all userConfig keys — gap noted, existing issue owns) / n/a (no config seam → not required).
The intro establishes SHIP 0 / STAY 14 / DROP 0 as the MCP-carry vocabulary. The discovery row reads perplexity/ref/microsoft-learn declared, not shipped (rule 3) — no SHIP/STAY/DROP label. Every other row uses the taxonomy or a dash. A consistent label like n/a — declared not shipped (rule 3) would align it.
Links to MIGRATION-PLAYBOOK.md have no fragment anchor. GitHub supports #heading-slug fragments on relative links; adding them (e.g., MIGRATION-PLAYBOOK.md#extensibility-contract-v21--the-four-seams) would drop readers at the right section rather than the top of a long document.
Summary
The document is accurate and the numerical claims are internally consistent. The Codex correctness finding was properly addressed before this re-review. All three remaining findings are editorial with no correctness impact. Ready to merge as-is; the three items above are nice-to-have polish.
The reason will be displayed to describe this comment to others. Learn more.
Stale comment
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole human repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
…issue
songwriting has 9 action skills + a songwriting/templates/ override seam and persists
output under CLAUDE_PROJECT_DIR — not knowledge-only. Moved from pure-reference to behavior
with a setup-action gap (parity with code-tidying's tidy-lanes); filed #1433. Counts:
pure-reference 5->4, behavior 27->28, net-new 5->6. Addresses Codex PR-review finding.
Refs melodic-software/medley#1388
Review docs/extensibility-contract-grading.md for accuracy and completeness
Check alignment with CLAUDE.md guidelines
Verify prior review findings are addressed
Post final review findings
Three commits in; the Codex songwriting finding has been addressed. Numerical claims are internally consistent after the reclassification. Three prior editorial findings remain open. No correctness or security concerns. Ready to merge.
Codex finding — addressed ✓
Commit 265938b reclassifies songwriting from pure-reference to behavior (9 action skills, template-override seam, ${CLAUDE_PROJECT_DIR}/songwriting/ output), emits setup issue melodic-software/medley#1433, and updates all counts accordingly. Finding resolved.
The Setup action dimension lists Present / missing / n/a (no config seam → not required) — three values. The table now uses five distinct values: present, bold **missing** (net-new gap), plain missing (existing tracked gap), partial, and n/a. A reader relying solely on the key will be confused. Suggested replacement for line 26:
Present / missing (net-new gap, issue filed this session) / missing (existing tracked gap, existing issue owns it) / partial (setup ships but does not cover all userConfig keys — gap noted, existing issue owns) / n/a (no config seam → not required).
The intro establishes SHIP 0 / STAY 14 / DROP 0 as the vocabulary for MCP-carry verdicts. The discovery row reads perplexity/ref/microsoft-learn declared, not shipped (rule 3) — no SHIP/STAY/DROP label. Every other row uses the taxonomy or a dash. A label like n/a — declared not shipped (rule 3) would make it consistent.
Both MIGRATION-PLAYBOOK.md links at L4–5 lack fragment anchors. GitHub resolves #heading-slug fragments on relative links — adding e.g. MIGRATION-PLAYBOOK.md#extensibility-contract-v21--the-four-seams would land readers at the relevant section instead of the top of a long document.
4. PR body description is now stale (informational)
The PR body still says "36 of 41 need no net-new work — 9 hook plugins, 5 pure-reference plugins" and "5 net-new setup-action gaps." After the songwriting reclassification these are 35/41, 4 pure-reference, and 6 gaps. No correctness impact to the shipped doc, but worth a body edit if you want the PR description to match.
Summary
The audit document is accurate after three corrective commits. Codex's songwriting finding is resolved. The three editorial findings above are carry-forwards with no correctness impact. Ready to merge as-is; findings 1–3 are nice-to-have polish before or after merge.
The reason will be displayed to describe this comment to others. Learn more.
Stale comment
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole human repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
…fy seam-vs-seam-3
work-items carries an optional .github/recurring-schedule.json tracked-config seam with only
per-item scaffolding and no dedicated setup; filed #1435. Added the config-seam vs seam-3
distinction so 'none needed' rows (reading the consumer's pre-existing CLAUDE.md / REVIEW.md /
tool configs) are defensibly separated from plugin-defined config seams. Net-new 6->7.
Addresses Codex PR-review finding.
Refs melodic-software/medley#1388
Check latest commit (8d800f1) against Codex work-items finding
Verify prior editorial findings status
Post final review findings
Fourth commit addresses the Codex work-items finding and adds a valuable seam-3 clarification. Numerical claims are internally consistent. Three prior editorial findings remain open. Ready to merge.
Codex finding — addressed ✓
Commit 8d800f1 adds the work-items seam-2 analysis: the row now shows **partial** setup, **GAP → net-new setup issue** verdict, and issue melodic-software/medley#1435 in the net-new table. The seam-3 clarification added to the Setup action dimension (lines 26–32) is an excellent addition — it clearly distinguishes plugin-defined config files (which require a setup action) from pre-existing consumer conventions read via seam 3 (which do not). This pre-empts similar misclassifications for review-toolkit, teach, claude-config-audit, etc.
The key ends with Present / missing / n/a (no config seam → not required). The table now uses six distinct values: Present, bold **missing** (net-new gap), plain missing (existing tracked gap), bold **partial** (net-new partial — work-items), plain partial (existing partial — machine-health), and n/a. The bold/plain convention is internally consistent and well-designed, but a reader relying solely on the key will not know how to decode partial or the bold distinction.
Suggested replacement for the key line:
Present / missing (net-new gap, issue filed this session) / missing (existing tracked gap, existing issue owns it) / partial (setup ships but is incomplete — net-new issue filed) / partial (existing partial gap, existing issue owns it) / n/a (no config seam → not required).
The intro establishes SHIP 0 / STAY 14 / DROP 0 as the vocabulary for MCP-carry verdicts. The discovery row reads perplexity/ref/microsoft-learn declared, not shipped (rule 3) — no SHIP/STAY/DROP label. Every other row uses the taxonomy or a dash. A label like n/a — declared not shipped (rule 3) would make it consistent.
Both MIGRATION-PLAYBOOK.md links lack fragment anchors. GitHub resolves #heading-slug fragments on relative links — adding e.g. MIGRATION-PLAYBOOK.md#extensibility-contract-v21--the-four-seams would land readers at the right section rather than the top of a long document.
Summary
All three Codex correctness findings have been addressed across four commits. The seam-3 clarification added in this latest commit is a meaningful improvement to the grading key. Numerical claims are internally consistent throughout. The three remaining findings above are all editorial with no correctness impact. Ready to merge; findings 1–3 are nice-to-have polish.
The reason will be displayed to describe this comment to others. Learn more.
Stale comment
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole human repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
Verify status of three open editorial findings from prior review
Identify new staleness issues from sibling PRs
Post final review findings
Merge commit dd042c6 synced to dc5aaf7 (PR #98 / review-toolkit), not to the current tip of main (6ad0148, PR #100 / implementation). Two rows in the grading doc are now factually stale against what will be on main at merge time. Three prior editorial findings remain open.
The row reads missing (plain, existing-tracked gap owned by #1420). PR #100 (6ad0148) is already on main and resolves #1420: it bumps implementation to v0.2.0 and ships /implementation:setup (the re-runnable ladder writer for .claude/ecosystems/*.yaml). The merge commit dd042c6 incorporated main only up to dc5aaf7; PR #100 landed after that point. When this PR merges into main, the grading doc will describe a setup gap that has already shipped.
The row ends "compliant; ecosystem-commands retrofit owned by #1421." PR #98 (dc5aaf7) is a parent of the merge commit dd042c6, so the ecosystem-commands retrofit is already in this branch's working tree: review-toolkit is at v0.2.0, ecosystem-specialist.md has the command-truth resolution ladder, and #1421 is resolved. The qualifier "owned by #1421" should be removed or marked as resolved.
Every other MCP-carry cell uses the established SHIP/STAY/DROP taxonomy or a dash. The discovery row reads perplexity/ref/microsoft-learn declared, not shipped (rule 3) — no taxonomy label. A label like n/a — declared not shipped (rule 3) would align it.
Both MIGRATION-PLAYBOOK.md links lack fragment anchors. GitHub resolves #heading-slug fragments on relative links; adding them (e.g. MIGRATION-PLAYBOOK.md#extensibility-contract-v21--the-four-seams) would land readers at the correct section.
Summary
The audit document has been accurate through four corrective commit rounds. The merge from main introduced a gap: the implementation row (finding 1) is factually wrong at merge time because PR #100 has already shipped /implementation:setup on main, and the review-toolkit row (finding 2) carries a stale outstanding-issue reference that was resolved by PR #98 before the merge commit. Both are one-line fixes. The three carry-forward items (3–5) remain editorial with no correctness impact. Recommend addressing findings 1–2 before merging.
The reason will be displayed to describe this comment to others. Learn more.
Stale comment
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole human repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
Verify prior correctness findings — both confirmed stale against merged state
Draft all corrections locally (committed as 1430c92) — push blocked by repo write permissions
The merge commit at 1981c5c pulled in PR #100 (6ad0148, implementation v0.2.0), which resolves the two correctness issues the previous review raised. I've prepared and committed the fixes locally but cannot push due to a 403 (write access not granted for this bot). The five changes below are ready to apply.
Correctness findings (pre-merge staleness — apply before merging)
The row still reads missing for Setup and "owned by #1420." PR #100 ships plugins/implementation/skills/setup/SKILL.md and bumps the plugin to v0.2.0. The config seam also expanded — /implementation:setup writes .claude/ecosystems/<eco>.yaml (seam 2, ecosystem-commands contract) in addition to userConfig notes_dir.
Every other row uses the taxonomy or a dash. Change perplexity/ref/microsoft-learn declared, not shipped (rule 3) to n/a — perplexity/ref/microsoft-learn declared, not shipped (rule 3).
Summary
All five Codex correctness findings (machine-health, songwriting, work-items) were addressed in prior commits. The two new correctness issues introduced when PR #100 merged into this branch have been prepared as commit 1430c92 locally — the changes are confirmed correct against plugins/implementation/skills/setup/SKILL.md and both plugin.json files (v0.2.0). Push was blocked by a 403; Kyle can cherry-pick or manually apply the three correctness changes (findings 1–3) before merging. Findings 4–5 remain editorial polish with no correctness impact.
The reason will be displayed to describe this comment to others. Learn more.
Risk: low. Did not approve because Cursor Bugbot completed with skip status (usage limit reached), so the required automated review did not complete. Human review is needed before merge; no reviewers were assigned because the PR author is the sole human repository contributor.
Sent by Cursor Approval Agent: Pull Request Router and Approver
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Grades all 41 shipped plugins against the extensibility contract v2.1 (four seams, setup-action requirement, convention-resolution ladder, baked-assumption check, MCP-carry, evals) and records the snapshot as a new audit doc
docs/extensibility-contract-grading.md, indexed from the README alongside the smoke-tests record.Findings
bug-report,discovery,planning,code-tidying,claude-opseach expose a config seam but ship no setup/configure action. One retrofit issue filed per plugin, sub-issue-linked under wave-2 map:Doc-only change (no
marketplace.json/ catalog-table touch) — conflict-free with sibling PRs.Refs melodic-software/medley#1388
Note
Low Risk
Documentation-only; no plugin code, marketplace catalog, or runtime behavior changes.
Overview
Adds
docs/extensibility-contract-grading.md, a dated audit snapshot that grades all 41 marketplace plugins against extensibility contract v2.1 (class, config seams, setup actions, convention ladder, baked assumptions, MCP-carry, evals).The doc concludes 34 plugins need no net-new work, flags 7 behavior plugins with missing or partial setup for their config seams (
bug-report,claude-ops,code-tidying,discovery,planning,songwriting,work-items), records 0 baked repo assumptions, and links seven new retrofit issues (#1428–#1435) under wave-2 #1369. It also notes reclassifyingsongwritingfrom pure-reference to behavior and defers eval backfill to #1396.README indexes the new grading doc next to the existing smoke-tests contract doc.
Reviewed by Cursor Bugbot for commit 1981c5c. Bugbot is set up for automated code reviews on this repo. Configure here.