Skip to content

feat(session-flow): gate optional plugin calls on enabled skills - #16450

Merged
kyle-sexton merged 19 commits into
mainfrom
feat/seam-gate-enabled
Nov 16, 2026
Merged

kyle-sexton merged 19 commits into
mainfrom
feat/seam-gate-enabled

Conversation

@kyle-sexton

@kyle-sexton kyle-sexton commented Nov 15, 2026 •

Copy link
Copy Markdown
Member

Closes #16430

Siblings: #16446

Summary

Completes item #16430 on top of #16446, which already landed the seam-phrasing contract change, the hook-unit rule, the handoff "Plugin profile" sentence, the course-digest edit and the handoff eval. This PR adds what #16446 left: the marketplace-wide sweep of gate sites from "installed" to "enabled", the remaining scan edits, the checks that follow the contract wording, and ADR 0109 recording the plugin-profiles one-way doors.

Fix

  • Gate sites swept to the enabled wording (classification below), plus docs/plugin-philosophy.md's rule line and the .claude/rules successor rule.
  • plugins/session-flow/reference/structure.md: the save-point's suggested-skills slot says "if enabled" (the "Plugin profile" sentence is feat(session-flow): gate optional plugin calls on available skills and record the plugin profile in handoffs #16446's).
  • Scan sites, re-read first: workflow/context/continuation.md gates /discovery:read-docs (fallback WebFetch); orient gates /visualization:visualize (fallback an inline mermaid block); planning/plan gates /discipline:pick-for-the-problem and /performance:goal with an inline fallback, and its ## Next drops "runs in this session", which a planning profile without the implementation plugin makes false.
  • Fallback and remedy clauses beside a gate say "not enabled" where they said "absent", and a remedy says to enable the plugin, installing it first when it is not installed. docs/contracts/loop-lane/README.md carries the same two wording changes.
  • Checks that follow the contract wording: plugin-quality's collect-standards.sh accepts an enabled gate and still grades a convention worded "explicit installed-ness"; skill-quality's ## Next check flags is|if|when enabled as whole words, each with a test.
  • Prose rewrites with no gate in them (context-usage, session-flow structure, prototype, user-experience, user-interface), which a fix-docs lane pass had added, are reverted: no required check asked for them, so those files differ from main only by their gate edits.
  • ADR docs/adr/0109-switch-plugin-sets-by-profile-and-gate-optional-calls-on-enabled-skills.md: profile format and keys, profile homes, extend rule, marketplace governance, the two hook units, core removal, the seam gate, the Phase 1 probe record, and the owner's Q75 answer (harness-ops does not depend on session-telemetry).
  • plugins/planning/tests/interview-defenses.test.sh: two section digests re-pinned for the gate wording in the planning interview's SKILL.md Step 4 and loop.md relentless mechanics; no defense sentence changed.
  • One changelog fragment per touched plugin.

Wording basis: the merged contract allows "an equivalent ... clause" and itself uses "when the other-plugin plugin is enabled" (docs/contracts/seam-phrasing/README.md, "The shape" item 1), so swept sites read "(if that plugin is enabled)" / "(if enabled)".

Gate-site classification

Swept every tracked markdown file for installed-ness phrasing (plugin is installed, (if installed), when installed, when it is installed, when X is (not) installed, skill/capability is installed) and edited gate sites only.

Not a gate (left unchanged) Reason
**/evals/**, **/fixtures/**, script tests fixtures: scenario text and sample answers
docs/adr/**, CHANGELOG.md historical records
reference/plugin-reconfiguration.md copies and their contract "the project the plugin is installed in": install location
artifact-protocol.md copies "assume another plugin is installed": the hard-dependency rule, vendored copies
harness-ops/skills/plugins/**, audit-skill-visibility, audit-mod-posture, render-report remedies plugin-sync and install-state wording
playgrounds/**, native-playground.md presence lines, installed from its marketplace cross-marketplace provenance check and its parity token
READMEs, docs/catalog.md, plugin.json and SKILL.md frontmatter descriptions descriptive; the contract puts the gate where the invocation is instructed
docs/upstream/**, vendor/**, skill-authoring's paraphrase of upstream upstream text
tool installs (jq, shellcheck, pwsh modules, a browser tool, Codex CLI) tool availability, not a plugin gate
whether or not ... is installed, (not installed) state labels, run-e2e's (app-control not installed) report string not a condition on a call

Verification

  • bash plugins/skill-quality/scripts/check-skill.test.sh: all assertions passed after merging main.
  • bash plugins/plugin-quality/skills/audit/scripts/collect-standards.test.sh: all passed.
  • uvx pytest plugins/session-flow/scripts/tests/test_save_point.py plugins/session-flow/scripts/tests/test_workflow_continuation_rule.py: 290 passed.
  • bash plugins/planning/tests/interview-defenses.test.sh: PASS=185 FAIL=0 on the merged head.
  • Pre-push check-affected: passed on the merged head.
  • Fallback eval: feat(session-flow): gate optional plugin calls on available skills and record the plugin profile in handoffs #16446's plugins/session-flow/evals/ho-disabled-profile-plugin-no-entry passed 4 of 4 runs (Gate optional plugin calls on enabled skills and record the profile in handoffs #16430 (comment)); this branch's own workflow eval was dropped as a duplicate.
  • Evals not run: per-skill evals for the other swept skills. The sweep swaps a word inside gate clauses and adds fallback and remedy text; it removes no instruction, and the eval gate on skill body edits covers a cut to an existing skill (docs/review/skills-and-agents.md, plugins/playbooks/skills/skill-authoring/reference/skill-criteria.md "A cut to an existing skill ships only when its evals show no loss").

Related

Refs #16428, #16446

Follow-ups

  • Pre-existing (advisory): no check fails when a gate site comes back worded "installed"; a check could scan tracked markdown for installed-ness gate phrasing with an allowlist built from the non-gate table above.

🤖 Generated with Claude Code

https://claude.ai/code/session_015XWwmd6RuLMb2eXqV5uoEJ

kyle-sexton and others added 2 commits November 15, 2026 16:36
The seam-phrasing gate now asks whether the target skill is among the
available skills (its plugin enabled), not whether the plugin is
installed, and every classified gate site says "enabled". The handoff
save-point writes the restore entry /plugin-profiles:profile status
prints when that skill is available. The hook-unit rule joins the hook
review rules, and ADR 0109 records the plugin-profiles one-way doors.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RAoAtUUc1niAWM9BoSeLXE
#16446 already landed the seam contract, hook-unit rule, handoff sentence and
handoff eval; take main's side there, drop this branch's duplicate hook rule and
its unrun workflow eval, and record the owner's Q75 answer in ADR 0109.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmd6RuLMb2eXqV5uoEJ
@kyle-sexton
kyle-sexton marked this pull request as ready for review November 16, 2026 00:25
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 16, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-16T00:28:01.286471Z 970d6de Draft marked ready
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-refine/fix-docs
Model rung 0
Leg handoff
Run job success
Activity step success
Commits pushed none (read lane)
Run 43780279009

Failed GitHub API calls

Method Path Status
GET /repos/{repo}/git/ref/tags/* 404

Lane-Activity: pr-refine/fix-docs
@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-refine/apply-patch
Model rung 0
Leg base
Run job success
Activity step success
Commits pushed ef2b04d
Run 43780279009

Failed GitHub API calls

Method Path Status
GET /repos/{repo}/git/ref/tags/* 404

@github-actions

github-actions Bot commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

claude-security-review reviewed df7973a55398f427d0296b758b2569ee45c27b5d: read 304 of 304 changed file(s); 0 finding(s), 0 as inline comments.

Chunk 1 of 7 (40 files)

Security review, chunk 1 of 7: no security issues found

I read every hunk of all 40 files in .claude-lane/chunks/chunk-1.patch. This lane already reviewed an earlier head of this pull request (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. 13 new changelog fragments under .changes/ and ADR 0109, which records the plugin-profile design and its probe results. The rest are prose edits in skill and reference files (adhd, ai-briefing, ai-slop, app-control, architecture, attribution, autonomy, bugs, code-metrics), docs/plugin-philosophy.md, docs/contracts/loop-lane/README.md with its CHANGELOG, and .claude/rules/skill-bodies-state-current-rules.md. Almost every hunk changes the condition on an optional call to another plugin from "installed" to "enabled". A plugin must be installed to be enabled, so each condition lets through the same cases or fewer. Every fallback stays in place.

Trusted paths. .claude/rules/skill-bodies-state-current-rules.md matches **/.claude/**, so I read its whole change and what it governs. "installed-ness gate" becomes "availability gate" and "installed" becomes "enabled". The rule only says what a ## Next section may contain. The edit widens nothing and removes no prohibition.

Instruction-surface lens. No hunk deletes, narrows or softens a standing guardrail. The loop-lane README edit changes only the condition for invoking the discipline sweep and for using its inline fallback; the fallback's instructions are unchanged. The security-defect routing in architecture/improve/actions/deepening.md (report privately, file nothing) is unchanged; only the condition on the non-security /bugs:write hand-off changed. I did not consult the protected-class register, because nothing was removed.

Trust controls and manifests. No file in this chunk touches a workflow, an action under .github/actions/, a merge-hold script, the token broker, the trusted-actor list, CODEOWNERS or a plugin manifest. No manifest lines apply.

Identical copies. identical.txt lists the adhd changelog fragment and 44 copies under .changes/<plugin>/. A fragment does the same thing in any plugin directory, so no copy is a finding.

Settled threads and prior findings. No settled thread covers a line in this chunk. The lane's one earlier post has no line anchor and reported no findings, so nothing needs restating.

Embedded text. No PR data tries to direct the lane. A closing-issue comment asks readers not to edit the issue bodies without asking first; that is process guidance, not an attempt to steer this review.

Spend. No billing change. The PR context mentions an eval run of about 4.3 USD on API credit, outside this diff; this chunk adds no spend.

Chunk 2 of 7 (40 files)

Security review, chunk 2 of 7: no security issues found

I read every hunk of all 40 files in .claude-lane/chunks/chunk-2.patch. This lane already reviewed an earlier head (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. Every hunk is prose in skills, agents and references across these plugins: code-metrics, code-tidying, codebase-health, context-budget, context-usage, corrections, coupling, debugging, developer-experience, discipline, discovery, disk-hygiene and docs-hygiene. Each one changes the condition for an optional call to another plugin from "installed" to "enabled". A plugin has to be installed before it can be enabled, so each gate now lets through the same cases or fewer, and every fallback stays as it was.

The added remedy text in four places prints claude plugin enable <plugin>@<marketplace>, with claude plugin install … before it only when the plugin is missing:

  • corrections/codify issues and preferences
  • developer-experience audit-tools, build-cli and setup

These commands are printed for the person to run. The skill does not run them, and "Name no tracker or forge CLI command" is kept.

Instruction-surface lens. No hunk deletes, narrows or softens a guardrail. In code-tidying/skills/tidy/SKILL.md, the fallback used when source-control is not enabled still keeps every PR-creation invariant inline:

  • issue-linkage resolution
  • body assembly through a quoted heredoc rather than an unquoted <<EOF
  • refusing gh pr create until the body has a closing keyword or the opt-out marker

Widening that fallback to the installed-but-disabled case therefore reaches the same safeguards. The batch-simplify agent ladder now says "available" where it said "installed" for the third rung. That rung chooses the agent type and grants no extra tools. I read docs/contracts/instruction-exception-register/README.md only by reference: its classes are prevention rules, and none of these hunks removes one.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. None touches a plugin manifest, a workflow, an action under .github/actions/, a merge-hold script, the token broker or the trusted-actor list, so no manifest lines apply.

Settled threads and prior findings. The settled thread on developer-experience/skills/audit-tools/SKILL.md (plus setup and build-cli) is fixed in this head: each site now prints the enable command. Its lines are unchanged since it was resolved, and I raise nothing new on it. The lane's only earlier post is a summary with no line anchor and no findings to repeat.

Embedded text. The PR context and the settled-threads file contain no text that tries to direct this lane.

Spend. No billing or cost change.

Chunk 3 of 7 (40 files)

Security review, chunk 3 of 7: no security issues found

I read every hunk of the 40 files in .claude-lane/chunks/chunk-3.patch. They are skill and reference prose in docs-hygiene, docs-naming, education, explainer-video, firecrawl, github, harness-config, harness-memory, harness-ops and implementation. The prior-findings file says blocking only: yes, so this review reports only CRITICAL and IMPORTANT findings. There are none.

What changed. Almost every hunk changes the condition on an optional call to another plugin from "installed" to "enabled". The matching fallback clauses change from "absent" or "not installed" to "not enabled". This matches the pull request's stated purpose. A plugin can be enabled only after it is installed, so each gate now lets through the same cases or fewer. Every fallback branch stays as it was. The only other edit is in implementation:implement-dispatch item 18, where a step reference changes from "Step 3's cadence" to "Step 2's cadence". That is a cross-reference fix with no security effect.

Instruction-surface lens. No hunk deletes, narrows or softens a guardrail. I checked the security-adjacent sites:

  • harness-config:unhobble Phase 4. Before the merge, it runs /review:security-review when the review plugin is enabled. Without the plugin, the pull request body must still record that the retired rules got no security pass. If the plugin is installed but disabled, the skill could not be invoked anyway, so no protected rule loses its check.
  • implement-dispatch. The phase-boundary review and the cross-vendor verifier keep their fresh-context fallbacks. Tracker-edge reads still STOP when the seam cannot read edges. The worker-side worktree rule still STOPs rather than editing the default checkout. Untrusted input still skips the workspace setup step.
  • harness-ops:changelog. When a stage's plugin is not enabled, its rows are still reported as not executed and are never done by hand. Filing still needs confirmation from the user.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. None touches a plugin manifest, a workflow, an action under .github/actions/, a merge-hold script, the token broker or the trusted-actor list. No manifest lines apply.

Settled threads and prior findings. The lane's only earlier post is a summary comment with no inline findings, so there is nothing to repeat.

Embedded text. The PR context includes an issue comment asking readers not to edit the issue bodies without first checking with another session. It is addressed to people and does not try to direct this lane. Nothing else in the PR data tries to direct the lane.

Spend. This chunk changes no billing or cost settings.

Chunk 4 of 7 (40 files)

Security review, chunk 4 of 7: no security issues found

I read every hunk of all 40 files in .claude-lane/chunks/chunk-4.patch. This lane already reviewed an earlier head (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. The chunk edits skill, context, reference and template prose in these plugins: implementation (implement and its mode contexts), improvement, instruction-placement, knowledge, mutation-testing, overengineering, performance, pixel-art, planning and playbooks. Almost every hunk changes the condition on an optional call to another plugin from "installed" to "enabled", or changes "absent" to "not enabled". A plugin must be installed before it can be enabled, so each gate now lets through the same cases or fewer. Every fallback stays in place.

Two edits add text:

  • planning/skills/plan/SKILL.md adds an inline fallback for a skill whose plugin is not enabled, and rewords a ## Next line.
  • knowledge/skills/course-digest/SKILL.md says that when testing is not enabled, Phase 1 uses the first browser tool the session offers and names it in the run report. Credential handling is unchanged: credentials still live in shell environment variables and the cookie store under ${CLAUDE_PLUGIN_DATA}. Nothing new reaches a credentialed path.

plugins/planning/tests/interview-defenses.test.sh changes two section digests and adds a comment for each. This is a test pin, not a security control. The pinned sections changed only in their gate clauses.

Instruction-surface lens. No hunk deletes, narrows or softens a guardrail. These rules are all kept: "never invent a skill name", "absence of a pipeline skill is never license to implement", the --exercised refusal without testing, "do not commit-and-push unilaterally", and the TDD rules. docs/contracts/instruction-exception-register/README.md was not read. That was not needed, because no rule was removed.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. None touches a plugin manifest, a workflow, an action under .github/actions/, a merge-hold script, the token broker or the trusted-actor list. No manifest lines apply.

Settled threads and prior findings. The settled threads on planning/skills/interview/context/handoff.md and the other files they name were fixed at earlier heads. Their lines are unchanged, and I raise nothing new on them. The lane's only earlier post is a summary comment with no inline findings, so nothing needs restating.

Embedded text. The PR data I read contains no text that tries to direct the lane. One closing-issue comment asks readers not to edit the issue bodies; it does not address this review.

Spend. No billing or cost change.

Chunk 5 of 7 (40 files)

Security review, chunk 5 of 7: no security issues found

I read every hunk of all 40 files in .claude-lane/chunks/chunk-5.patch. This lane already reviewed an earlier head (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. Most hunks are prose in skills and reference files under playbooks, playwright, plugin-quality, prototype, repo-fleet-hygiene, repo-hygiene, review (quality-gate and change-set) and session-flow. They change the condition on an optional call to another plugin from "installed" to "enabled". A plugin can only be enabled once it is installed, so each gate now lets through the same cases or fewer, and every fallback is unchanged.

Three hunks add text:

  • orient gets an inline mermaid fallback.
  • review/reference/change-set.md and quality-gate/context/security.md now suggest enabling claude-security@claude-plugins-official, installing it first if needed. These are recommendations to the person; the skills do not run them.

Code changes.

  • plugin-quality/skills/audit/scripts/collect-standards.sh first looks for the needle "explicit availability condition" and falls back to the older "explicit installed-ness". The convention path comes from the existing topic_rel. The match uses grep -n -F with quoted arguments in cite, so no shell or regex injection is added. The awk window now also accepts enabled as a gate word. This only affects an advisory probe that reports possible convention gaps, not a security control. The test file adds a matching case.
  • session-flow/scripts/save_point.py only changes a string literal.

Instruction-surface lens. No hunk deletes, narrows or softens a guardrail. No protected-class rule is removed: mutation gates, "never report that check as passed", "Emit no porcelain-based substitute verdict", the worktree no-delete rule and the loosened-check spec rules are all unchanged.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. None touches a plugin manifest, a workflow, an action under .github/actions/, a merge-hold script, the token broker or the trusted-actor list, so no manifest lines apply.

Settled threads and prior findings. The lane's only earlier post is a summary comment with no inline findings. Nothing in this chunk repeats a prior or settled point.

Embedded text. The PR context includes a commenter's request not to edit issue bodies without asking a named session first. It is aimed at people, not at this lane, and does not try to direct this review.

Spend. No billing or cost change.

Chunk 6 of 7 (40 files)

Security review, chunk 6 of 7: no security issues found

I read every hunk of all 40 files in .claude-lane/chunks/chunk-6.patch. This lane already reviewed an earlier head (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. These are skill and reference prose edits in session-flow, songwriting, source-control, testing, toolchain, user-experience, user-interface and verification. They also touch two testing eval files and the check-skill.sh ## Next check in skill-quality, along with its tests. Almost every hunk changes the condition on an optional call to another plugin from "installed" (or "absent") to "enabled" (or "not enabled"). A plugin must be installed before it can be enabled, so each gate now lets through the same cases or fewer. Every fallback stays in place.

Instruction-surface lens. No guardrail is deleted, narrowed or softened:

  • source-control/babysit-prs/reference/safety.md:1030: the autopilot approval pass still needs a genuine review. When no review plugin is enabled, it falls back to an equivalent thorough manual review. The gate still requires the approval to exist on the live head.
  • pull-request/reference/ready-for-review.md:382: the security review still runs. When the review plugin is not enabled, it falls back to a security-reviewer agent or a fresh-context subagent.
  • pull-request/reference/comments-ci.md:77: the manifest dependencies: flag settle rule now names an "if enabled" check instead of "if installed". The rule is otherwise unchanged: it still records INCORRECT only when the plugin invokes the dependency with no gate, condition, fallback or negation.
  • testing/run-e2e: the skill still "installs and enables nothing". The install or enable step is a suggestion that the person carries out.
  • verification/confirm: Stage 1 still stops on failure, and both the e2e step and the live-app fallback still run when their plugins are not enabled.

Script change. plugins/skill-quality/scripts/check-skill.sh:2476 adds a fixed ERE alternative and a sed trim. Both run over $NEXT_BLOCK, a here-string of the skill's own text. Nothing reaches a shell or eval, so there is no injection path.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. The babysit-prs files changed here are under reference/, not the trusted scripts/ modules. The chunk touches no plugin manifest, workflow, action under .github/actions/, merge-hold script, token broker or trusted-actor list. No manifest lines apply.

Settled threads and prior findings. No settled thread or prior inline finding concerns these lines.

Embedded text. None of the PR data I read tries to direct the lane.

Spend. None in this chunk.

Chunk 7 of 7 (20 files)

Security review (chunk 7 of 7): no security issues found

I read every hunk of all 20 files in this chunk: visualization (visualize SKILL.md and decision-matrix), wizard/generate, the work-items references and skills (attend-queue, decompose and its contexts, onboard-adapter, setup and providers, ship and actions/run.md, track add and done, triage, work), and the two prompts/loops/ lane prompt files. This lane already reviewed an earlier head (blocking only: yes), so this pass reports only CRITICAL and IMPORTANT findings. There are none.

What changed. Every hunk changes the condition on an optional call to another plugin from "installed" (or "absent") to "enabled" (or "not enabled"). A plugin must be installed before it can be enabled, so each gate now lets through the same cases or fewer. Every fallback is unchanged. work-items/skills/ship/actions/run.md also adds a remedy for the user: enable the implementation plugin, installing it first if needed. The skill still stops before any claim. This matches the PR's stated purpose.

Instruction-surface lens. No hunk deletes, narrows or softens a guardrail. In work-items/skills/work/SKILL.md, these rules are all unchanged apart from the gate wording:

  • the strong-tier model binding when implementation is not enabled;
  • the human-merge hold read through merge-hold.sh;
  • disabling auto-merge before a push, and the rule that no lane arms auto-merge;
  • the lane-push.sh push identity;
  • skipping workspace setup for an item marked provenance: untrusted.

In wizard/generate, browser driving still needs a per-row yes from the user and an isolated profile, and the agent still reads no secrets. In the attended-queue lane prompts, only the planning gate wording changes.

Trust controls and manifests. No file in this chunk matches a trusted-path glob. None touches a plugin manifest, a workflow, an action under .github/actions/, a merge-hold script, the token broker or the trusted-actor list, so no manifest lines apply.

Settled threads and prior findings. No prior inline finding falls on these lines. The lane's earlier summary comment reported no findings, so nothing needs restating.

Embedded text. None of the PR data I read tries to direct the lane.

Spend. This chunk changes no billing, cost or run settings.

critical-findings: 0

@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/resolve-conflicts
Model rung 0
Leg handoff
Run job success
Activity step success
Commits pushed none (read lane)
Run 43781803268

Failed GitHub API calls

Method Path Status
GET /repos/{repo}/git/ref/tags/* 404

@github-actions github-actions Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

claude-review: 6 inline finding(s) on ef2b04d5d177e0dc8fcfc6bb885c873dcf79419f; the summary is in this lane's pull request comment.

claude-lane: claude-review

@github-actions

github-actions Bot commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

claude-review reviewed df7973a55398f427d0296b758b2569ee45c27b5d: read 304 of 304 changed file(s); 0 finding(s), 0 as inline comments.

Chunk 1 of 7 (40 files)

claude-review, chunk 1 of 7: no blocking findings. I read every hunk of all 40 files in this chunk. This lane already reviewed earlier heads (blocking only: yes), so this pass reports blocking findings only.

What this chunk changes

  • Changelog fragments. 13 new fragments, all bump: patch. That level fits a change that only swaps gate wording and fallback/remedy text.
  • Gate wording. Cross-plugin gates change from "installed" to "enabled" across adhd, ai-briefing, ai-slop, app-control, architecture, attribution, autonomy, bugs and code-metrics. The same change is made in docs/plugin-philosophy.md, the loop-lane contract (with CHANGELOG 19.7.1) and .claude/rules/skill-bodies-state-current-rules.md. In that rules file, the ## Next ban now names "availability gate".
  • ADR 0109. New. Its links to ADR 0075, docs/review/hooks.md and seam-phrasing/README.md#the-shape all resolve. The hook-unit rule it cites is in docs/review/hooks.md:7-9, which meets the closing issue's "the hook-unit rule is in the hook review rules". No other ADR uses the number 0109.

Closing issue #16430. For the files in this chunk, "every classified gate site uses the enabled wording" holds. Each gate is either rewritten or, for ADRs and changelogs, exempt as a historical record. The PR body's test and eval results are the author's claims; I did not run them.

Settled and prior findings. None of the six resolved threads cover a file in this chunk. I have nothing to post again.

Advisory, left out because this pass is blocking only: in drift-delta-sweep.md:64-92, the three lane gates now say "enabled", but their fallbacks still say "where it is absent". The PR body says fallback clauses next to a gate were changed from "absent" to "not enabled". Here "absent" can still be read as "not among the available skills", so behaviour does not change.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane. One issue comment asks people not to edit the issue bodies again without asking the control-hub session first. It is addressed to people, not to this lane, and it changes nothing here.

Spend and hold. Nothing in this chunk can cause a billed charge, and it has no critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 2 of 7 (40 files)

claude-review, chunk 2 of 7: no blocking findings. I read every hunk of all 40 files in this chunk. This lane already reviewed earlier heads of this pull request, so this pass reports blocking findings only.

What changed. Gate clauses in code-metrics, code-tidying, codebase-health, context-budget, context-usage, corrections, coupling, debugging, developer-experience, discipline, discovery, disk-hygiene and docs-hygiene now say "enabled" instead of "installed". The routes in corrections codify (issues, preferences) and in developer-experience audit-tools, build-cli and setup now print claude plugin enable <plugin>@<marketplace>, with claude plugin install … before it only when the plugin is not installed.

What I checked

  • Fallbacks match their gates. Each remedy that names a fix in this chunk matches the "enabled" gate. That includes the four developer-experience sites from the earlier settled thread (audit-tools/SKILL.md:81-84, setup/SKILL.md:75-78 and :151-153, build-cli/SKILL.md:21-25). None of these lines changed since that thread was resolved, so it stays settled.
  • No gate left half-changed. None of the edited sentences still mixes "installed" and "enabled" in a way that leaves a disabled plugin without a path. batch-simplify's agent ladder now reads "any other available agent", which is consistent with the rest of it.
  • Changelog fragments. Every plugin this chunk touches has a fragment: .changes/<plugin>/feat-seam-gate-enabled-*.md is in files.txt for all 13 of them.
  • Closing issue Gate optional plugin calls on enabled skills and record the profile in handoffs #16430. For the files in this chunk, the criterion "every classified gate site uses the enabled wording" holds. The PR body's test and eval results are the author's claims; I did not run them.

Advisory, left out because this pass is blocking only. A few unchanged lines next to a changed gate still say "absent" where the PR body says those fallback lines now say "not enabled":

  • plugins/codebase-health/skills/audit/templates/checklist.md:254 ("When those plugins are absent")
  • plugins/context-budget/skills/audit/reference/report.md:289 ("absent it")
  • the heading at plugins/discovery/skills/explore/reference/ecosystem-discovery.md:36

The fallback still works in each case, so none of these blocks the merge.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane.

Spend and hold. This chunk only changes skill prose, so it can cause no billed charge and contains no critical security issue. There is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 3 of 7 (40 files)

claude-review, chunk 3 of 7: no blocking findings. I read every hunk of all 40 files in this chunk: docs-hygiene, docs-naming, education, explainer-video, firecrawl, github, harness-config, harness-memory, harness-ops and implementation implement-dispatch. This lane has already reviewed earlier heads, so this pass reports blocking findings only.

What changed

  • Gate wording. Gates on optional plugins now say "enabled" instead of "installed". Their fallback clauses say "not enabled" in most places, and the remedies name the plugin to enable.
  • Step number. implement-dispatch item 18 now points the test-first cadence at "/implementation:implement Step 2" instead of Step 3. I checked this against plugins/implementation/skills/implement/SKILL.md. Step 2 is "Execute with Incremental Validation", where the cadence lives. Step 3 is "Divergence Detection". The new reference is correct.

What I checked

  • Fallbacks match their gates. Earlier rounds found this class of problem. In this chunk, the sites from earlier threads are fixed:
    • audit-native-overlap report strings now say "skill-quality not enabled".
    • lane-catalog.md branch label is now Not enabled:.
    • changelog/SKILL.md now says "naming the plugin to enable".
    • decisions.md now says "With the plugin not enabled".
    • In item-list.md, the tracker gate and its STOP clause now agree.
  • Closing issue Gate optional plugin calls on enabled skills and record the profile in handoffs #16430. For the files in this chunk, "every classified gate site uses the enabled wording" holds. The PR body's test and eval results are the author's claims; I did not run them.

Advisory, left out because this pass is blocking only. A few fallbacks still say "absent" under a gate that now says "enabled":

The PR's Fix list says these clauses become "not enabled". Either way, the fallback still runs when the skill isn't listed.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane. One issue comment asks people not to edit the issue bodies again. It is addressed to people, not to this lane, and changes nothing here.

Spend and hold. This chunk is wording-only skill prose. Nothing here can cause a billed charge or is a critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 4 of 7 (40 files)

claude-review, chunk 4 of 7: no blocking findings. I read every hunk of all 40 files in this chunk: implementation implement and its contexts, improvement, instruction-placement, knowledge, mutation-testing, overengineering, performance, pixel-art, planning (including interview-defenses.test.sh), and the playbooks fable-5 and repo-sweep hygiene catalog. This lane already reviewed earlier heads (blocking only: yes), so this pass reports blocking findings only.

What changed

  • Gate clauses changed from "installed" to "enabled", and fallback clauses from "absent" or "not installed" to "not enabled".
  • planning/plan: /discipline:pick-for-the-problem and /performance:goal are now gated "(if enabled)". A new sentence says a skill that is not enabled is listed as unavailable and its concern is covered inline. The ## Next line no longer says "runs in this session".

What I checked

  • Gates and fallbacks match. This was the class of problem found in earlier rounds. In this chunk, each fallback beside a changed gate now says "not enabled", or names a remedy that fits a plugin that is installed but disabled. That includes planning/interview/context/handoff.md, whose earlier thread is settled and unchanged.
  • Test re-pins. The two re-pinned digests in interview-defenses.test.sh cover the SKILL.md Step 4 section and loop.md's "Relentless me mode mechanics" section, which runs through "Incremental persistence + branch-out". Both sections contain only the gate-clause edits their comments name. The other files this chunk touches, handoff.md, scenario-comparison.md and templates/checklist.md, are outside every digested region, so no other pin moves. I could not recompute the SHA-256 values without a shell. The PR body says the suite passed (PASS=185 FAIL=0); that is the author's claim and I did not verify it.
  • Closing issue Gate optional plugin calls on enabled skills and record the profile in handoffs #16430. For this chunk's files, "every classified gate site uses the enabled wording" holds. The plan-skill edit the lifecycle scan asked for is made as the PR describes.

Settled threads and prior findings. The handoff.md pointer thread is settled and the file now carries the suggested fix. This chunk repeats no prior finding.

Advisory, left out because this pass is blocking only: the overengineering audit and realign neighbor tables still head their fallback column "Inline fallback when absent". routing-out.md renamed the same kind of column to "not enabled".

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane.

Spend and hold. Nothing in this chunk can cause a billed charge or is a critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 5 of 7 (40 files)

claude-review, chunk 5 of 7: no blocking findings. I read every hunk of all 40 files in this chunk. Earlier heads have already been reviewed (blocking only: yes), so this pass reports blocking findings only.

What this chunk changes

  • Gate wording. Gates move from "installed" to "enabled" across playbooks (repo-sweep, skill-authoring), playwright demo-video, plugin-quality (auditor, config.md, audit and its references), prototype, repo-fleet-hygiene, repo-hygiene, review (change-set.md, quality-gate and its contexts) and session-flow.
  • Scan edit. orient zoom-out now gates /visualization:visualize, with an inline mermaid block as the fallback. This is the scan edit the closing issue asks for.
  • Collector. collect-standards.sh cites "explicit availability condition" first and falls back to "explicit installed-ness". Its three-line window now also accepts enabled.

What I checked

  • Earlier findings are fixed at head. These threads are settled, and I confirmed the fixes in this chunk's diff:
    • repo-sweep/reference/next.md: both the gate and the pointer now say "not enabled".
    • rewrite-a-plugin.md and skill-criteria.md: both say "enabling that plugin would run it".
    • native-playground.md: both routes are gated on "enabled".
  • Remedies match their gates. review/reference/change-set.md and quality-gate/context/security.md now say to enable claude-security, installing it first when it is not installed.
  • Collector needle. docs/contracts/seam-phrasing/README.md:13 contains "explicit availability condition", so the new needle resolves against the real contract. The new test plants a home still worded "installed-ness" and checks that it is still graded and not reported as basis-missing.
  • Handoff fixtures. The save_point.py hint string changed. The session-flow test fixtures still contain "(if installed)" lines, but they are handoff fixtures, which the PR's classification table exempts. Nothing I found pins the hint's old text.
  • Author's claims. The PR body's test results (collect-standards.test.sh, test_save_point.py) are the author's claims. I did not run them.

Closing issue #16430. For the files in this chunk, "every classified gate site uses the enabled wording" holds. The orient scan edit is made.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane.

Spend and hold. Nothing in this chunk can cause a billed charge or is a critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 6 of 7 (40 files)

claude-review, chunk 6 of 7: no blocking findings. I read every hunk of all 40 files in this chunk. This lane already reviewed earlier heads, so this pass reports blocking findings only.

What changed

  • Gate wording. Optional cross-plugin calls now check "enabled" instead of "installed", across session-flow, songwriting, source-control, testing, toolchain, user-experience, user-interface and verification.
  • Scan site. workflow/context/continuation.md now gates /discovery:read-docs and falls back to WebFetch.
  • check-skill.sh. The ## Next check also flags is|if|when enabled as whole words, with two new tests.

What I checked

  • The new ## Next pattern. (if that plugin is enabled) matches is enabled), which the sed trims to is enabled, as the test expects. In "this enabled", the h before is fails the word boundary, so nothing matches.
  • Fallbacks match their gates. The run-e2e remedies now say "enable, installing it first if it is not installed" at Step 1, the Suggestion: line, Step 4 and ## Next. That fixes the last pass's advisory. verification:confirm's Step 4 note matches. The eval run-e2e-route-install-suggestion now expects "installing and enabling", which matches the skill.
  • Strings left unchanged. The (app-control not installed) report strings are in the PR's non-gate table.
  • Closing issue Gate optional plugin calls on enabled skills and record the profile in handoffs #16430. For this chunk, "every classified gate site uses the enabled wording" holds. The PR's test results are the author's claims; I did not run them.

Advisory, left out because this pass is blocking only: the PR says fallbacks next to a gate now say "not enabled" where they said "absent". Two still say "when it is absent" under an "enabled" gate:

The fallback probably still runs for a disabled plugin, but the PR changed the same clause in testing/skills/diagnose/context/loop.md and testing/skills/write/SKILL.md, so the wording is inconsistent.

Settled and prior threads. None touch this chunk's files.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane.

Spend and hold. Nothing here can cause a billed charge or is a critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

Chunk 7 of 7 (20 files)

claude-review, chunk 7 of 7: no blocking findings. I read every hunk of the 20 files in this chunk: visualization visualize (SKILL.md, decision-matrix.md), wizard generate, the work-items references and skills (execution-shape.md, pipeline-shape.md, attend-queue, decompose and its two context files, onboard-adapter, setup and providers.md, ship and actions/run.md, track add.md and done.md, triage, work), and the two loop-lane prompt files. This lane has already reviewed earlier heads, so this pass reports blocking findings only.

What changed. Every hunk changes a gate clause from "installed" to "enabled", for example "when the writing plugin is enabled" and "(if enabled)". Two hunks also change the matching fallback: ship Step 4 now says "when a named plugin is not enabled", and work Step 5 now says "When the implementation plugin is not enabled" and "when that plugin is not enabled".

What I checked

  • Fallbacks match their gates. This was the class of problem found in earlier rounds. No fallback in this chunk tells the user to install a plugin under an enabled gate.
    • ship/actions/run.md:27 now says to enable implementation, installing it first if needed. That resolves the earlier thread, which the owner settled.
    • The visualize playground routing gates both the playground skill and the playgrounds wrapper on "enabled". That matches the settled native-playground.md fix.
  • Non-gate wording. Sentences such as "work-items installs standalone/independently of planning" (pipeline-shape.md, both loop-lane prompts) describe how a plugin is packaged. They are not conditions on a call, so leaving them unchanged fits the PR's classification table.
  • Prompt parity. prompts/loops/loop-lane-profile-claude-code-plugins.md and prompts/loops/loop-lane-prompts.md get the same one-line edit, so the two copies still match.
  • Closing issue Gate optional plugin calls on enabled skills and record the profile in handoffs #16430. For this chunk, "every classified gate site uses the enabled wording; non-gate sites are unchanged" holds. The PR body's test and eval results are the author's claims; I did not run them.

Embedded text. The PR context, settled threads and prior findings contain no instructions aimed at this lane. The issue comment asking that the issue body not be edited without checking first is addressed to people editing the issue, not to this lane.

Spend and hold. Nothing in this chunk can cause a billed charge or is a critical security issue, so there is no hold.

Verifier dropped 0 of 0 candidates.

critical-findings: 0

Lane-Activity: pr-unblock/resolve-conflicts
@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/apply-conflict-patch
Model rung 0
Leg base
Run job success
Activity step success
Commits pushed 2f30321
Run 43781803268

Failed GitHub API calls

Method Path Status
GET /repos/{repo}/git/ref/tags/* 404

kyle-sexton and others added 2 commits November 15, 2026 19:57
…r a disabled plugin

The sweep moved gates to "enabled" but left some fallbacks and remedies
saying "not installed" or "install it". For a plugin that is installed but
switched off, those fallbacks either never fired or told the user to
install something already installed.

- Fallback conditions now read "not enabled" (handoff.md, next.md,
  lane-catalog.md, native-playground.md, scope-budget.md, find).
- Report strings say "not enabled" (audit-native-overlap, changelog).
- Remedies say to enable the plugin, installing it first only when it is
  not installed (audit-tools, setup, build-cli, ship run, confirm,
  rewrite-a-plugin, skill-criteria).
- Gates the sweep missed in files it already touched now read "enabled"
  (visualize write-css, code-metrics size.md, songwriting research).

Refs: #16430

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmd6RuLMb2eXqV5uoEJ
@kyle-sexton

Copy link
Copy Markdown
Member Author

Reply to the claude-review summary finding from chunk 6, finding 1 (the gate site the sweep missed in visualization/skills/visualize/SKILL.md):

CORRECT. Basis: plugins/visualization/skills/visualize/SKILL.md:94 still read "and /user-interface:write-css is installed, use it ...; without it, write the CSS here". That is an invocation gate with a fallback, two lines below the chart-craft sentence this PR changed.

Fixed in 3c32a39. It now reads "is enabled".

The same commit fixes the rest of this class across the PR diff, beyond the inline threads:

  • missed gates in files the sweep already touched, now "enabled": code-metrics/skills/audit/actions/size.md:85,88,90, code-tidying/skills/tidy/reference/scope-budget.md:85, improvement/skills/find/SKILL.md:147,150, songwriting/skills/suno/context/research-recipes.md:162-163, session-flow/skills/retro/reference/ecosystem-improvement-catalog.md:33
  • a remedy that told the user to install: verification/skills/confirm/SKILL.md:172 now names enabling app-control, installing it first if it is not installed.

kyle-sexton and others added 4 commits November 15, 2026 21:12
The seam-phrasing sweep changed one gate clause in each of two pinned
interview sections, "(if installed)" to "(if enabled)" on the
/session-flow:workflow reference. No defense sentence changed. The two
section digests are re-pinned with plugins/planning/tests/repin.sh.

Refs: #16430

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
… plugins

/corrections:codify issues still gated on the work-items plugin being
installed, so with work-items installed but off it called
/work-items:track and never reached its fallback. It and its preferences
sibling now gate on enabled and print `claude plugin enable`, preceded by
`claude plugin install` when the plugin is not installed.

A second sweep over every sibling of a swept file and every tracked
markdown file under plugins/ found eight more gate sites still worded as
installed: ai-briefing's slide fallback routing, discipline
use-your-skills, implement-dispatch's tracker edges, plugin-quality's
role seams, session-flow audit-friction and keep-going, and testing
run-e2e's own-driving condition and Next remedy.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
Decision 5 rests on the owner's Q75 answer, which the Refs line's
Q55 to Q70 range did not cover.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
@kyle-sexton

Copy link
Copy Markdown
Member Author

Disposition for the claude-review finding on 7d48c47 (plugins/corrections/skills/codify/actions/issues.md still gated on "installed").

CORRECT, fixed in 98f6686. Basis: issues.md:7-9 read "Use it when the work-items plugin is installed", so with work-items installed but off the action passed the gate and never reached its fallback. It now gates on enabled and prints claude plugin enable work-items@<marketplace>, preceded by claude plugin install when the plugin is not installed. The sibling preferences.md gate, already on "enabled", gets the same remedy.

Because this was a second missed site, the tree was swept again: every sibling file of a swept gate (134 directories) and all tracked markdown under plugins/, classified against the body's gate-site table. Eight more gate sites moved to the enabled wording in the same commit: ai-briefing slide-generation, discipline use-your-skills, implementation item-list, plugin-quality audit, session-flow audit-friction and keep-going, testing run-e2e and its own-driving context. Remaining hits are the table's non-gate classes (fixtures and evals, changelogs, READMEs, reconfiguration and artifact-protocol copies, descriptions, state labels, tool installs).

Advisory (ADR 0109 Refs line omitted Q75): fixed in 510a9cd.

Checks on 510a9cd: plugins/planning/tests/interview-defenses.test.sh PASS=185 FAIL=0 with no pin moved, plugins/planning/scripts/repin-digests.test.sh 23 passed, pre-push check-affected passed.

@github-actions

Copy link
Copy Markdown
Contributor

The pr-verify verifier failed head 510a9cd88dc4a938c623bf399b7d5772695e2b69. What it wrote, as data:

What it wrote (71 lines)
Summary: Whole spec, fail: Issue #16430 (closed by PR #16450; #16446 is a credited sibling). Most of
  the spec is delivered, but two done-when items are delivered by no chunk and are not excused, so
  the head fails. What is covered: AC14 is dropped because #16446 plainly says it delivered the
  gated `/plugin-profiles:profile status` sentence and the write-nothing fallback, and chunk 2
  confirms it added no second sentence. The seam-phrasing contract change and the hook-unit rule
  also come from #16446. Chunks 1, 2 and 3 each say the installed-to-enabled sweep is done for their
  own files, and together those files cover the manifest's gate-site edits; chunk 1 also found no
  missed gate site in its parts. Chunk 2 delivers all four lifecycle-scan sites (continuation,
  orient, plan, course-digest SKILL.md), and #16446 did course-digest workflow.md. Chunk 1 delivers
  ADR 0109, which records every one-way door the issue names. Not covered: the fallback-eval item.
  #16446 added the eval but says it never ran i
Chunk 1 of 3, pass-notes: Chunk 1 of 3 for PR #16450, which closes #16430 with #16446 credited as a
  sibling. I read all 18 parts and both manifest parts whole. Every file in this chunk is text;
  there are no binary files. Each done-when item, as far as this chunk shows:

(1) AC14, the one gated `/plugin-profiles:profile status` sentence in the handoff structure
  reference: #16446 delivered it (credited sibling). None of this chunk's files touch it.

(2) The seam-phrasing contract dropping "if that plugin is installed": delivered by #16446. The rest
  of item 2, the marketplace-wide gate-site sweep, is the work #16446 left open. This chunk delivers
  that sweep for its own files: `.claude/rules/skill-bodies-state-current-rules.md`,
  `docs/plugin-philosophy.md`, and every plugin file from adhd through
  `implementation/skills/implement-dispatch/reference/item-list.md`. Each optional cross-plugin gate
  changes from installed to enabled wording, and each fallback is kept. Where a remedy used to pri
Chunk 2 of 3, pass-notes: Chunk 2 of 3 for #16430 (closes; sibling #16446 credited). I read all 20
  parts of chunk 2 whole, plus both manifest parts, context, spec, spec-text, upstream jobs and the
  capture summary. The capture is `blocked` (credit-stop), which is no evidence either way.

What these parts deliver:
- **Gate-site sweep, partly.** "Every classified gate site uses the enabled wording; non-gate sites
  are unchanged" is delivered for every file in this chunk, from implementation/implement through
  work-items/triage. Each invocation gate on another plugin's skill moves from installed to enabled
  with its fallback kept. Install remedies now say to enable the plugin, installing it first if
  needed (work-items ship/run.md, testing run-e2e, verification confirm). The non-gate text I
  spot-checked was left alone, which matches the PR's classification table: the `(app-control not
  installed)` state label, "prefer installed skills", "installable plugin" in native-playground.md,
  and "needs 
Chunk 3 of 3, pass: Chunk 3 of 3 for #16430 (closes; #16446 is a credited sibling). This chunk has
  three files, all text-only gate-wording changes and no binary files. (1)
  plugins/work-items/skills/work/SKILL.md: three optional-plugin gates change from "installed" to
  "enabled": `/app-control:drive` "when enabled" in the before_after_runnable condition, the
  `source-control` plugin's worktree creation seam "when the source-control plugin is enabled" with
  a plain `git worktree add` fallback, and the fix worker dispatched as `implementation:implementer`
  "when the implementation plugin is enabled", whose existing explicit-model fallback still applies
  when the plugin is absent. Apart from those words, the long rewritten lines match the originals.
  Gates already worded "among the available skills" and the "Without the source-control plugin" /
  "When the implementation plugin is absent" fallback clauses are correctly left as they were. (2)
  prompts/loops/loop-lane-profile-claude-code-plugins.md 
- Done-when item 'The fallback eval passes within its run budget of 4 runs': no chunk delivers it,
  and #16446 says the eval was not run. The 4/4 pass sits only in an issue comment on #16430. Record
  that run as a `person_checks` entry in #16430's refinement, or deliver it through a credited
  sibling, so the item is excused.
- Done-when item 'AC7 (process criterion of the effort): carried by the container's PR': no chunk
  delivers it, and neither `person_checks` nor a credited sibling excuses it. Record it as a person
  check, or move it out of #16430's done-when list into the container issue #16428.
- plugins/context-usage/reference/reader-contract.md:564: outside the spec: "widely-cited
  practitioner anchor" becomes "practitioner anchor". This is a wording edit with no gate in it, so
  it is not part of the gate sweep #16430 asks for. It is harmless, but a person should confirm it
  is meant to ride along in this PR.
- plugins/session-flow/reference/structure.md:248 and :528: wording rewrites unrelated to gates (the
  'machinery reads as mission' sentence and 'This is the section that beats compaction. Write it
  long.'). The spec did not ask for them; a human should confirm the meaning is unchanged.
- plugins/prototype/context/discipline.md: the 'No polish' row and the 'mock before you wire'
  ordering note were reworded. This is outside the #16430 gate sweep.
- plugins/user-experience/reference/flows-ia.md and
  plugins/user-interface/reference/types/web/marketing-pages.md: several prose sentences were
  rewritten with no gate in them (error-page, tour, motion, hero-replay and launch-day lines). This
  is outside the #16430 scope.
- plugins/skill-quality/scripts/check-skill.sh:2476: `(is|if|when) enabled` has no word boundary, so
  text like 'this enabled' inside a `## Next` block would also raise the operative-chain warning. It
  only warns, but adding `\b` would stop false positives.

This pull request leaves the merge queue, and ci-status holds it out while it carries pr-verify-failed. A pr-verify pass removes the label, at a new commit or at this same head: once a trusted person edits the body, and only trusted people have edited it since the fail, this head is judged again, so a fail on the body or the issues it links needs only the body fixed, no new commit.

@github-actions github-actions Bot added the pr-verify-failed pr-verify failed this head; a pass at a new commit, or after a trusted body edit, removes it. label Nov 16, 2026
kyle-sexton and others added 5 commits November 15, 2026 22:16
This reverts commit ef2b04d. Its edits
reword prose with no gate in it; no required check asked for them.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
…d-ness

A convention adopted before the availability-condition wording now gets
the seam check instead of a basis-missing finding.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
…dy states

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
"this enabled" no longer reads as an "is enabled" gate. The boundary is
a POSIX bracket class, since the shell-portability check bans \b.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
kyle-sexton and others added 4 commits November 15, 2026 22:29
The test-first brief line pointed at Step 3, which is divergence
detection.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
…oute

The route suggestion names enabling app-control, installing it first
when it is not installed, matching its ## Next. The report string
"(app-control not installed)" stays: seven eval cases grade it.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
…gins

Gates still worded as installed (a named agent, the routing check, a
skill catalog, owning skills, the loop-lane discipline preamble) say
enabled, fallbacks that said absent say not enabled, and the
claude-security remedy names enabling it, installing first if needed.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015XWwmcNG8zpLhdZzJdZw5y
@kyle-sexton

Copy link
Copy Markdown
Member Author

Dispositions for the pr-verify fail note and both pr-review-main verdicts on 510a9cd. Fixes are on df7973a (commits d957fd4, 44a147f, 92d2081, 96ab0a5, 3f91aaa, f83cb1f, 59d9c5d, df7973a, on a merge of main).

pr-verify

Item Classification Basis and change
Done-when "fallback eval passes within 4 runs" delivered by no chunk Spec fixed A run is not a file change. #16430 now lists it under "Checked outside the pull request" with the 4 of 4 run linked. A person-checks record could not be used: only the lanes App's refinement comment counts as one.
Done-when "AC7" delivered by no chunk Spec fixed Moved out of #16430's done-when; it is the container's criterion and #16428's Brief already lists it.
Prose rewrites with no gate (context-usage reader-contract, session-flow structure, prototype discipline, user-experience flows-ia, user-interface marketing-pages) CORRECT, reverted in d957fd4 They came from a fix-docs lane commit, a model refine pass and not a required check; no check script names those phrases. Those files now differ from main only by gate edits.

pr-review-main: claude-review and claude-security-review

Finding Classification Basis and change
docs/plugin-philosophy.md "prefer an installed named agent" CORRECT, fixed in 59d9c5d Now "enabled"; the nearby "is absent" fallback now reads "is not enabled".
collect-standards.sh accepted only the new phrase, so a convention on the old wording lost the check CORRECT, fixed in 44a147f It falls back to "explicit installed-ness" when the new phrase is missing; a new old-wording test fails without the fallback. The contract holds the new phrase word for word: docs/contracts/seam-phrasing/README.md "The gate is an explicit availability condition on the invocation".
session-flow workflow SKILL.md "check what the consuming setup actually installs"; work-items pipeline-shape.md CORRECT, fixed in 59d9c5d Both now speak of enabled skills.
Cross-chunk: does check-skill.sh's regex match the rule text, and does the philosophy gate line match what collect-standards.sh accepts Confirmed, plus boundary fix in 96ab0a5 The rule's ## Next paragraph and the check's warning agree, and the philosophy line's "if enabled" is accepted. The pattern now matches whole words; a test shows "this enabled" no longer warns. The repository's shell-portability gate bans \b, so the POSIX boundary form is used.
work-items work SKILL.md fallbacks say "absent"; run-e2e remedy says only "install" CORRECT, fixed in 59d9c5d and f83cb1f Fallbacks say "not enabled"; the remedy says enable, installing first if needed, and the eval case that grades that suggestion is updated with it. The "(app-control not installed)" report string stays: seven eval cases grade it, and the body's table classes it as a state label.
interview-defenses.test.sh re-pin comment names one of two edits CORRECT, fixed in df7973a The comment names the gate clause and its Pointer fallback. Comment only; no pin moved (PASS=185 FAIL=0).
Changelog fragments against each plugin's diff; implement-dispatch's step citations PARTIAL, fixed in 3f91aaa and 59d9c5d 57 touched plugins, 57 fragments, none missing either way. One citation was wrong ("Step 3's cadence" is Step 2) and is fixed.
SUGGESTION: no check fails when an installed gate comes back Not built here Listed under ## Follow-ups in the body, since advisory issue filing is paused in this repository.
Security lane's two suggestions Confirmed, no change The cross-checks it asked for were made; nothing to fix.

Sibling sweep, third pass: every candidate was opened by hand. 24 more sites moved to the enabled wording or the enable-then-install remedy, across architecture, work-items, plugin-quality, overengineering, source-control, implementation, planning, knowledge, review, testing and the loop-lane contract. Left as not gates: tool and agent installs, install-state text, descriptions, reconfiguration copies, plugin-sync text, and the standing "prefer installed skills" instruction.

Checks on the new head: interview-defenses.test.sh PASS=185 FAIL=0, repin-digests.test.sh 23 passed, check-skill.test.sh and collect-standards.test.sh all passed, pre-push check-affected passed.

@github-actions github-actions Bot removed the pr-verify-failed pr-verify failed this head; a pass at a new commit, or after a trusted body edit, removes it. label Nov 16, 2026
@kyle-sexton
kyle-sexton added this pull request to the merge queue Nov 16, 2026
Merged via the queue into main with commit 3b27875 Nov 16, 2026
75 checks passed
@kyle-sexton
kyle-sexton deleted the feat/seam-gate-enabled branch November 16, 2026 04:03
@github-actions github-actions Bot added the merge-path: owner-agent Post-merge accounting record (README.md, Merge paths) label Nov 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

merge-path: owner-agent Post-merge accounting record (README.md, Merge paths)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Gate optional plugin calls on enabled skills and record the profile in handoffs

1 participant