feat: close the setup-contract wave — final setups + contract gate

plugins/disk-hygiene/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "disk-hygiene",
-  "version": "0.3.0",
+  "version": "0.4.0",
   "description": "Context-aware disk hygiene for arbitrary directory trees: inventories orphaned and temporary artifacts, classifies evidence into review tiers, and offers exact-path cleanup only after a fresh safety preview and explicit per-tier approval. The target is read-only by default; OS-managed paths, links and mount points, VCS-tracked content, changed entries, and live-handle uncertainty fail closed.",
   "author": {
     "name": "Melodic Software",

plugins/disk-hygiene/CHANGELOG.md

@@ -3,6 +3,18 @@
 All notable changes to the `disk-hygiene` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.4.0]
+
+### Added
+
+- **`/disk-hygiene:setup` skill on the uniform contract** (fleet conformance
+  wave, dim 8). `check` reads the clean skill's bundled scripts as the source
+  of truth and probes Python 3.11+, conditional Git, the current OS family's
+  documented lane (Linux `lsof` and macOS audit-only reported as INFO), and
+  the effective `disk_hygiene_enabled` toggle. `apply` is guidance-only with
+  no write path; toggle guidance states `--config`'s fresh-install-only
+  semantics. A disabled toggle downgrades prerequisite FAILs to INFO.
+
 ## [0.3.0]
 
 Fixes driven by a live Windows user-profile audit where the engine was unusable through its

plugins/disk-hygiene/README.md

@@ -57,6 +57,9 @@ at preview. Backups remain the recovery boundary for user data.
 - macOS supports audit/report only because this implementation has no authoritative bind-mount and
   descriptor-anchoring proof for its execution lane.
 
+Verify this machine's prerequisites and platform posture with `/disk-hygiene:setup check`;
+`/disk-hygiene:setup apply` resolves anything the check reports with guidance.
+
 ## Usage
 
 ```text

plugins/disk-hygiene/skills/setup/SKILL.md

@@ -0,0 +1,68 @@
+---
+name: setup
+description: "Verify the disk-hygiene plugin's runtime prerequisites and platform posture for this machine. Use when: 'set up disk-hygiene', 'configure disk-hygiene', 'is disk-hygiene working', a clean run reported a missing prerequisite, or before a first audit on a new machine. Actions: check (read-only verification, default) | apply (resolve what check found). Re-runnable and safe."
+argument-hint: "check | apply"
+user-invocable: true
+disable-model-invocation: true
+---
+
+## Purpose
+
+Thin check-centric setup per the uniform contract: `check` inspects and reports, `apply`
+resolves. This plugin owns no consumer-project configuration — targets and modes arrive as
+`/disk-hygiene:clean` arguments, and the only tunable is the native `userConfig` toggle —
+so `apply` is pure guidance and writes nothing.
+
+Action routing: no argument or `check` runs the check; `apply` runs the check first, then
+points at each remediation. Both are non-interactive — never prompt when the action is given.
+
+## `check` (read-only)
+
+The clean skill and its bundled scripts (`${CLAUDE_PLUGIN_ROOT}/skills/clean/`) are the
+single source of truth for what the plugin requires per platform. **Read them first** —
+probe what they actually require, don't recite this file. Then run each probe via Bash and
+report a PASS/FAIL/INFO table with one remediation line per FAIL.
+
+When the plugin's toggle is disabled, every prerequisite absence downgrades from FAIL to
+INFO — a deliberately disabled plugin is not broken. Report the probes informationally and
+note that re-enabling restores the FAIL semantics.
+
+1. **Python 3.11+ on `PATH`** — the interpreter used by scanning, validation, the
+   skill-scoped guard, and cleanup. FAIL if absent or older, with the README's requirement
+   as the remediation; the plugin never downloads a runtime. Report the absolute
+   interpreter path (guarded engine calls must use the same absolute interpreter the guard
+   reports — Bash aliases and functions cannot substitute).
+2. **Git** — `command -v git`. Conditional per the README: optional for ordinary trees,
+   required when a target contains or sits inside a Git worktree. Report presence as INFO
+   with that conditionality stated; absence is only a FAIL for worktree-containing targets.
+3. **Platform posture** — detect the current OS family and report its documented lane per
+   the README: Windows (full, `lstat` reparse + Win32, never UAC), Linux (full when
+   `/proc/self/mountinfo` is readable; `lsof` needed only for the optional execution
+   lane — absent `lsof` is INFO with the reduced-capability note), macOS (audit/report
+   only by design — INFO, not a defect).
+4. **Hook toggle** — report the effective `disk_hygiene_enabled` value:
+   `${user_config.disk_hygiene_enabled}` (unexpanded or empty means default `true`).
+5. **Plugin registration** — INFO: confirm the plugin is enabled for this project
+   (`/plugin` → Installed) rather than parsing settings files.
+
+## `apply` (idempotent)
+
+Run `check`, then for each FAIL point at the resolution. Every prerequisite is a system
+tool or an OS capability, so `apply` installs nothing and writes nothing — it only points:
+
+- missing/old Python: the platform's own Python 3.11+ install channel; never a plugin
+  download.
+- missing git (worktree targets): platform install instructions.
+- toggle off: direct to `/plugin configure disk-hygiene` (interactive, any time).
+  Headless: `--config` only applies on a fresh install (ignored once installed), so
+  reconfigure via `claude plugin uninstall disk-hygiene` then
+  `claude plugin install disk-hygiene@<marketplace> --config disk_hygiene_enabled=true`;
+  this skill never writes user settings or `pluginConfigs`.
+
+Re-running `apply` after everything passes changes nothing and reports "already configured".
+
+## What this skill does NOT do
+
+- Run an audit or cleanup — that is `/disk-hygiene:clean`.
+- Write the plugin cache, Claude Code user settings, or `pluginConfigs`.
+- Install any tool or runtime, during either `check` or `apply` — guidance only.

plugins/guardrails/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "guardrails",
-  "version": "0.6.2",
+  "version": "0.7.0",
   "description": "Eight safety guards that block secret/credential writes, hardcoded machine-specific paths, git hook-bypass attempts, irreversible git operations (force-push, reset --hard, worktree-wide checkout/restore discards), Bash file-write workarounds that circumvent Write/Edit hooks, (advisory) hallucinated CLI flags, (advisory) un-throttled Workflow fan-out that risks burst 529s, and (advisory) direct git commit/gh pr create calls bypassing this marketplace's own commit/pull-request skills — each independently toggleable.",
   "author": {
     "name": "Melodic Software",

plugins/guardrails/CHANGELOG.md

@@ -3,6 +3,19 @@
 All notable changes to the `guardrails` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.7.0]
+
+### Added
+
+- **`/guardrails:setup` skill on the uniform contract** (fleet conformance
+  wave, dim 8). `check` reads the guard scripts and `hooks.json` as the
+  source of truth and probes Bash 5.0+, `jq` (absence = every guard fails
+  open — surfaced as the FAIL it is), each guard's effective toggle, the
+  `cli-flag-verify` scan surface, and the `block-dangerous-git` allowlist.
+  `apply` is guidance-only with no write path; reconfiguration guidance
+  states `--config`'s fresh-install-only semantics. All-toggles-disabled
+  downgrades prerequisite FAILs to INFO.
+
 ## [0.6.2]
 
 ### Changed

plugins/guardrails/README.md

@@ -146,6 +146,10 @@ as before.
 /plugin install guardrails@melodic-software
 ```
 
+Then verify the runtime prerequisites and live guard surface with
+`/guardrails:setup check`; `/guardrails:setup apply` resolves anything the
+check reports with guidance.
+
 ## License
 
 MIT (SPDX-License-Identifier: MIT). See the `LICENSE` file at the root of the

plugins/guardrails/skills/setup/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: setup
+description: "Verify the guardrails hooks' runtime prerequisites and per-guard toggle state for this machine. Use when: 'set up guardrails', 'configure guardrails', 'is guardrails working', 'which guards are on', a guard failed open with a jq notice, or after tuning guard toggles. Actions: check (read-only verification, default) | apply (resolve what check found). Re-runnable and safe."
+argument-hint: "check | apply"
+user-invocable: true
+disable-model-invocation: true
+---
+
+## Purpose
+
+Thin check-centric setup per the uniform contract: `check` inspects and reports, `apply`
+resolves. This plugin owns no consumer-project configuration — every tunable is a native
+`userConfig` option (eight per-guard toggles plus the `cli_flag_verify_bins`,
+`cli_flag_verify_skip_bins`, and `block_dangerous_git_allow` scalars) — so `apply` is pure
+guidance and writes nothing.
+
+Action routing: no argument or `check` runs the check; `apply` runs the check first, then
+points at each remediation. Both are non-interactive — never prompt when the action is given.
+
+## `check` (read-only)
+
+The guard scripts (`${CLAUDE_PLUGIN_ROOT}/hooks/*.sh`) and `hooks.json` are the single
+source of truth for the guard inventory and each guard's runtime needs. **Read them
+first** — probe what they actually require, don't recite this file. Then run each probe
+via Bash and report a PASS/FAIL/INFO table with one remediation line per FAIL.
+
+When every guard's toggle is disabled, prerequisite absences downgrade from FAIL to INFO —
+a deliberately disabled plugin is not broken. Report the probes informationally and note
+that re-enabling restores the FAIL semantics.
+
+1. **Bash 5.0+** — the guards' documented runtime floor (Git Bash on native Windows).
+   FAIL below the floor with the README Requirements remediation.
+2. **`jq`** — `command -v jq`. FAIL if absent: per the README, every guard then fails
+   OPEN (disabled) with a one-line stderr notice — the machine is unguarded, which is
+   exactly what this check exists to surface.
+3. **Per-guard toggles** — report each guard's effective value from its
+   `${user_config.<guard>_enabled}` rendering (unexpanded or empty means default `true`),
+   one row per guard, so the user sees the live guard surface at a glance.
+4. **`cli-flag-verify` scan surface** — report the effective `cli_flag_verify_bins` /
+   `cli_flag_verify_skip_bins` values and INFO-note the guard's own behavior for scanned
+   binaries missing from `PATH` (skipped, never flagged — per the guard source).
+5. **`block-dangerous-git` allowlist** — report the effective `block_dangerous_git_allow`
+   value (patterns only, verbatim; it contains no secrets by design).
+6. **Hook registration** — INFO: confirm the plugin is enabled for this project
+   (`/plugin` → Installed) rather than parsing settings files.
+
+## `apply` (idempotent)
+
+Run `check`, then for each FAIL point at the resolution. Every prerequisite is a system
+tool and every tunable is native `userConfig`, so `apply` installs nothing and writes
+nothing — it only points:
+
+- missing `jq` / old Bash: platform install instructions from the README Requirements
+  section; this skill never installs system packages.
+- any toggle or scalar change: direct to `/plugin configure guardrails` (interactive, any
+  time). Headless: `--config` only applies on a fresh install (ignored once installed), so
+  reconfigure via `claude plugin uninstall guardrails` then
+  `claude plugin install guardrails@<marketplace> --config KEY=VALUE …` (repeatable);
+  this skill never writes user settings or `pluginConfigs`.
+
+Re-running `apply` after everything passes changes nothing and reports "already configured".
+
+## What this skill does NOT do
+
+- Exercise a guard — any matching tool call does that end-to-end.
+- Write the plugin cache, Claude Code user settings, or `pluginConfigs`.
+- Install any tool, during either `check` or `apply` — guidance only.
+- Weaken a guard: it reports and routes; disabling is always the user's explicit act
+  through the native configuration surface.

plugins/review/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "review",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "Code-review toolkit: six read-only reviewer agents (code, security, architecture, doc drift, build/test/lint, CI-log audit) plus two orchestration skills — a single-lens quality gate and a multi-surface review fan-out with severity-ranked, deduplicated findings.",
   "author": {
     "name": "Melodic Software",

plugins/review/CHANGELOG.md

@@ -3,6 +3,19 @@
 All notable changes to the `review` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.14.0]
+
+### Changed
+
+- **Setup adopts the uniform check/apply contract** (fleet conformance wave,
+  dim 8 — caught by the new contract gate rather than the wave list). `check`
+  runs the standards-contract binding's state-reading procedure read-only
+  (index presence, row-path validation, version delta) and reports; `apply`
+  carries the existing bootstrap/reconfigure/migration flow with its
+  explicit-confirmation gates intact, re-verifying after every write. The
+  by-reference discipline is unchanged — the procedure still lives in the
+  contract binding, not restated here.
+
 ## [0.13.0]
 
 ### Changed

plugins/review/skills/setup/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: setup
-description: "Configure the review plugin for this repository: bootstrap the consumer's standards index per the standards convention — the index review criteria resolve through — persisting docs/standards/ and, on relocation, .claude/standards.yaml. Use when: 'set up review', 'configure the review plugin', 'review setup', 'set up standards', 'bootstrap the standards index', or a review skill reports a missing or version-skewed standards index. Re-runnable — safe to invoke again to reconfigure or migrate."
-argument-hint: "(no arguments — interactive interview)"
+description: "Configure the review plugin for this repository: bootstrap the consumer's standards index per the standards convention — the index review criteria resolve through — persisting docs/standards/ and, on relocation, .claude/standards.yaml. Use when: 'set up review', 'configure the review plugin', 'review setup', 'set up standards', 'bootstrap the standards index', or a review skill reports a missing or version-skewed standards index. Actions: check (read-only verification, default) | apply (bootstrap, reconfigure, or migrate). Re-runnable."
+argument-hint: "check | apply"
 user-invocable: true
 disable-model-invocation: true
 ---
@@ -20,7 +20,16 @@ implement it by reference, do not restate it.
 Idempotent: re-running reads the current state and offers an update rather than overwriting blind;
 a re-run against a conforming, current-version index proposes no changes.
 
-## Task
+Action routing per the uniform contract: no argument or `check` runs the binding's state-reading
+procedure read-only and reports — index presence and resolved standards root, per-row path
+validation, and the DIRECTIONAL version delta — as a PASS/FAIL/INFO table with one remediation
+line per FAIL, writing nothing. `apply` runs `check` first, then the binding's bootstrap /
+reconfigure / migration flow below; after any write it re-runs the relevant probe and reports the
+actual result. Non-interactive when the state admits exactly one conforming action (the
+conforming-index short-circuit); the binding's explicit-confirmation gates (hand-authored README
+conversion, bootstrap writes) remain explicit user decisions, never silent.
+
+## `apply` task
 
 Plugin-side notes on top of the binding's procedure:
 

scripts/validate-plugin-contracts.mjs

@@ -39,6 +39,19 @@ for (const path of setupSkills) {
   if (!/^disable-model-invocation:\s*true\s*$/m.test(frontmatter)) {
     fail(path, "setup skills must set disable-model-invocation: true");
   }
+  // Uniform contract shape (PLUGIN-PHILOSOPHY "Setup is explicit and repeatable"):
+  // check is the default read-only action; apply exists unless the skill declares the
+  // userConfig-only check-only carve-out the doctrine sanctions.
+  if (!/^argument-hint:\s*"check(?:\s*\||\s*\[|")/m.test(frontmatter)) {
+    fail(path, 'setup skills must declare check as the leading action in argument-hint ("check", "check | apply ...", or "check [<subaction>]")');
+  }
+  const body = content.slice(content.indexOf("---", 3) + 3);
+  if (!/`check`/.test(body)) {
+    fail(path, "setup skills must document the read-only check action");
+  }
+  if (!/`apply`/.test(body) && !/check-only/i.test(body)) {
+    fail(path, "setup skills must document apply, or declare the check-only userConfig-only carve-out");
+  }
 }
 
 const setupContractFiles = pluginFiles.filter(