Add the Claude Code plugin marketplace; fix silently-broken skill frontmatter

.claude-plugin/marketplace.json

@@ -0,0 +1,33 @@
+{
+  "name": "agents-shipgate",
+  "owner": {
+    "name": "Three Moons Lab",
+    "email": "help@threemoonslab.com"
+  },
+  "description": "Agents Shipgate — the deterministic merge gate for AI-generated agent capability changes. Skill-only plugin: workflows for shipgate check, agents-shipgate verify, and shipgate audit --host; the scanner runs through the agents-shipgate CLI installed locally.",
+  "plugins": [
+    {
+      "name": "agents-shipgate",
+      "source": "./plugins/claude-code",
+      "description": "Run Agents Shipgate Tool-Use Readiness workflows from Claude Code: detect whether Shipgate applies, bootstrap the advisory gate, verify agent-capability PRs, and triage findings without fabricating human approval. Requires the agents-shipgate CLI (pipx install agents-shipgate, contract v10 or newer).",
+      "version": "0.14.0",
+      "author": {
+        "name": "Three Moons Lab",
+        "url": "https://threemoonslab.com/"
+      },
+      "homepage": "https://threemoonslab.com/",
+      "repository": "https://github.com/ThreeMoonsLab/agents-shipgate",
+      "license": "Apache-2.0",
+      "keywords": [
+        "agents",
+        "claude-code",
+        "tool-use-readiness",
+        "release-gate",
+        "mcp-security",
+        "policy-as-code"
+      ],
+      "category": "security",
+      "strict": true
+    }
+  ]
+}

.claude/commands/shipgate.md

@@ -1,5 +1,5 @@
 ---
-description: Run the prominent Agents Shipgate flows: check, verify, or audit --host
+description: "Run the prominent Agents Shipgate flows: check, verify, or audit --host"
 ---
 
 Arguments: `$ARGUMENTS`

CHANGELOG.md

@@ -2,6 +2,21 @@
 
 ## Unreleased
 
+- **Claude Code plugin marketplace.** The repo now doubles as a Claude Code
+  plugin marketplace (`/plugin marketplace add ThreeMoonsLab/agents-shipgate`,
+  then `/plugin install agents-shipgate@agents-shipgate`) — the symmetric
+  counterpart of the existing Codex marketplace. The plugin is skill-only
+  (the auto-triggering skill + the namespaced `/agents-shipgate:shipgate`
+  command); the scanner stays in the separately installed CLI and hooks stay
+  on the explicit `install-hooks` path. Byte-identity with the canonical
+  skill/command sources is test-pinned. Fixed in passing (caught by
+  `claude plugin validate`): the canonical `SKILL.md` and `/shipgate`
+  command shipped YAML frontmatter with an unquoted `:` in `description`,
+  which Claude Code loads as silently-empty metadata — breaking
+  description-based skill auto-triggering for every existing install. Both
+  are now quoted, all byte-identical copies synced, and a regression test
+  parses the frontmatter.
+
 - **Contract v10 (additive): machine-readable `verify_required` on the Codex
   boundary result.** `shipgate check` already escalated to `warn` and routed
   to `verify` when a diff touched a tool surface it cannot gate; that deferral

README.md

@@ -311,6 +311,20 @@ agent mode auto-enables, so a zero-flag `agents-shipgate verify` prints the
 compact agent result. Slash command, skill internals, and manual paths:
 [`docs/agents/use-with-claude-code.md`](docs/agents/use-with-claude-code.md).
 
+Prefer a plugin over a committed kit? This repo is also a Claude Code plugin
+marketplace — the skill-only symmetric counterpart of the Codex plugin below
+(workflows, not the scanner binary; install the CLI separately):
+
+```bash
+/plugin marketplace add ThreeMoonsLab/agents-shipgate
+/plugin install agents-shipgate@agents-shipgate
+```
+
+The plugin ships the auto-triggering `agents-shipgate` skill and the
+`/agents-shipgate:shipgate` command (plugin commands are namespaced). It does
+not ship hooks — install those explicitly with `agents-shipgate install-hooks
+--target claude-code --write`, which requires the CLI on `PATH`.
+
 **Codex** — install the skill-only plugin from this repo's marketplace, or
 write the repo-scoped kit directly:
 

adoption-kits/claude-code-skill/.agents-shipgate-kit-metadata.json

@@ -19,7 +19,8 @@
       "398e88622bf73b524f91405ffc5dbccde651c6a9cb7c2df035ab01d39a964e4f",
       "9860b9246057289450b425daa212f248be8082327101f2a5e6a355a266f779c1",
       "5062b30ee84b3871c6532ce0e6c7ad41ac0203a1bd29be51ac3342e01fcd09cb",
-      "6031a900efa636677786ceda881af0aa4630500683ee1845f4dd1439fdbe2e70"
+      "6031a900efa636677786ceda881af0aa4630500683ee1845f4dd1439fdbe2e70",
+      "1b4470b625110acdc6f1e3930b7e6e6f43653c493e9a44c7ccde0dd9c88b09bb"
     ],
     "prompts/add-shipgate-to-repo.md": [
       "ea3c37cfbbd42c40d164abfe21d468a3a5550d5384125f94a53c947dea6b4b2a",

adoption-kits/claude-code-skill/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: agents-shipgate
-description: Run prominent Agents Shipgate flows when a change touches what an AI agent can do: `shipgate check`, `agents-shipgate verify`, or `shipgate audit --host`. Use after adding or modifying MCP servers or tools, tool/function definitions (@tool, @function_tool), OpenAPI specs that describe agent tools, agent prompts, permission scopes, approval or confirmation policies, agent CI workflows, or shipgate.yaml — and before creating a PR for any such change. Also use to verify agent-related PRs, fix or triage Shipgate findings, add Shipgate to CI, or interpret Shipgate verifier/report artifacts. Triggers on phrases like "add shipgate", "verify this agent PR", "merge verdict", "release readiness for my agent", "tool-use readiness", "shipgate check", "agents-shipgate verify", "audit host grants", "shipgate.yaml", "agents-shipgate-reports/verifier.json", "agents-shipgate-reports/report.json", "fix shipgate finding".
+description: 'Run prominent Agents Shipgate flows when a change touches what an AI agent can do: `shipgate check`, `agents-shipgate verify`, or `shipgate audit --host`. Use after adding or modifying MCP servers or tools, tool/function definitions (@tool, @function_tool), OpenAPI specs that describe agent tools, agent prompts, permission scopes, approval or confirmation policies, agent CI workflows, or shipgate.yaml — and before creating a PR for any such change. Also use to verify agent-related PRs, fix or triage Shipgate findings, add Shipgate to CI, or interpret Shipgate verifier/report artifacts. Triggers on phrases like "add shipgate", "verify this agent PR", "merge verdict", "release readiness for my agent", "tool-use readiness", "shipgate check", "agents-shipgate verify", "audit host grants", "shipgate.yaml", "agents-shipgate-reports/verifier.json", "agents-shipgate-reports/report.json", "fix shipgate finding".'
 ---
 
 # agents-shipgate skill

docs/agents/use-with-claude-code.md

@@ -35,6 +35,27 @@ when those files exist:
 agents-shipgate init --workspace . --write --claude-code
 ```
 
+### Plugin marketplace (no committed files)
+
+Prefer not to commit skill files into the repo? This repository is a Claude
+Code plugin marketplace; the plugin is **skill-only** — it supplies the same
+auto-triggering skill plus the slash command, and updates with `/plugin
+update` instead of re-running `init`:
+
+```text
+/plugin marketplace add ThreeMoonsLab/agents-shipgate
+/plugin install agents-shipgate@agents-shipgate
+```
+
+Plugin commands are namespaced: the command installs as
+`/agents-shipgate:shipgate` (the committed-kit path keeps plain `/shipgate`).
+The plugin does not ship hooks or the scanner — install the CLI (`pipx
+install agents-shipgate`, contract v10 or newer) and add hooks explicitly
+with `agents-shipgate install-hooks --target claude-code --write`. To
+pre-provision the marketplace for a whole team, add it to
+`.claude/settings.json` under `extraKnownMarketplaces` and enable
+`"agents-shipgate@agents-shipgate"` in `enabledPlugins`.
+
 To install the surfaces manually (no CLI), from the root of the project where
 you want `/shipgate` and the skill available:
 

plugins/claude-code/.claude-plugin/plugin.json

@@ -0,0 +1,21 @@
+{
+  "name": "agents-shipgate",
+  "version": "0.14.0",
+  "description": "Run Agents Shipgate Tool-Use Readiness workflows from Claude Code. Skill-only: the plugin supplies the agents-shipgate skill and the /agents-shipgate:shipgate command; the scanner runs through the agents-shipgate CLI installed in the local environment (pipx install agents-shipgate, contract v10 or newer). For the deterministic hooks (trigger check after edits, full verify at Stop), run: agents-shipgate install-hooks --target claude-code --write.",
+  "author": {
+    "name": "Three Moons Lab",
+    "email": "help@threemoonslab.com",
+    "url": "https://threemoonslab.com/"
+  },
+  "homepage": "https://threemoonslab.com/",
+  "repository": "https://github.com/ThreeMoonsLab/agents-shipgate",
+  "license": "Apache-2.0",
+  "keywords": [
+    "agents",
+    "claude-code",
+    "tool-use-readiness",
+    "release-gate",
+    "mcp-security",
+    "policy-as-code"
+  ]
+}

plugins/claude-code/commands/shipgate.md

@@ -0,0 +1,81 @@
+---
+description: "Run the prominent Agents Shipgate flows: check, verify, or audit --host"
+---
+
+Arguments: `$ARGUMENTS`
+
+Agents Shipgate is the deterministic merge gate for AI-generated agent
+capability changes.
+
+If the arguments include `audit`, run the host-grant audit flow. If they include
+`check`, run the local boundary check. Otherwise run the verifier flow. The
+supporting adoption and scan commands remain available, but this slash command
+should lead with only the prominent flows: `shipgate check`, `agents-shipgate verify`,
+and `shipgate audit --host`.
+
+The canonical, self-contained verifier instructions live in the bundled prompt
+files. For verifier runs, read `prompts/verify-agent-diff.md`. Try these paths
+in order; use the first that exists:
+
+1. `.claude/skills/agents-shipgate/prompts/<recipe>.md` — bundled with the `agents-shipgate` skill if installed in this project.
+2. `prompts/<recipe>.md` — present when this repo is a clone of `agents-shipgate` itself.
+3. `https://raw.githubusercontent.com/ThreeMoonsLab/agents-shipgate/main/prompts/<recipe>.md` — last-resort fetch.
+
+Prominent commands:
+
+```bash
+AGENTS_SHIPGATE_AGENT_MODE=1 shipgate check \
+  --agent claude-code --workspace . --format codex-boundary-json
+AGENTS_SHIPGATE_AGENT_MODE=1 agents-shipgate verify \
+  --workspace . --config shipgate.yaml \
+  --base origin/main --head HEAD \
+  --ci-mode advisory --format json
+AGENTS_SHIPGATE_AGENT_MODE=1 shipgate audit --host \
+  --json --out agents-shipgate-reports/host-grants.json
+```
+
+For local uncommitted work, omit `--base`/`--head` so uncommitted edits are
+scanned. For committed PR/CI refs, make the base ref available first because
+`verify` never fetches.
+
+Required behavior (do not skip):
+
+1. Set `AGENTS_SHIPGATE_AGENT_MODE=1` for every CLI call so errors emit a `next_action` JSON line on stderr.
+2. Run `agents-shipgate contract --json` when available and use it to verify the installed CLI's schema versions and gating signal.
+3. For verifier runs, parse `agents-shipgate-reports/agent-handoff.json` first,
+   then `verifier.json`, `verify-run.json`, and
+   `report.json.release_decision.decision` as the release gate.
+4. For check runs, parse stdout as `shipgate.codex_boundary_result/v1` and
+   switch on `decision`, `completion_allowed`, `must_stop`, `first_next_action`,
+   `human_review`, and `repair`.
+5. For host audits, parse `agents-shipgate-reports/host-grants.json` when
+   `--out` is used, or stdout when running JSON-only.
+6. Do **not** bypass the verifier by suppressing findings, lowering severity,
+   expanding baselines or waivers, removing Shipgate CI, or weakening agent
+   instructions. Verify-mode `SHIP-VERIFY-*` checks route those trust-root edits
+   to human review.
+7. Add `agents-shipgate-reports/` to `.gitignore` if it is not already.
+
+Report back: `release_decision.decision` and `reason`, `merge_verdict`,
+`can_merge_without_human`, blocker / review-item counts, top 3 active findings
+by severity, `verifier_summary` trust-root flags when present, capability
+change highlights, and one suggested next step.
+
+## Ongoing PRs
+
+For an ongoing PR that changes agent tools, MCP exports, OpenAPI specs, prompts,
+permissions, policies, CI gates, or `shipgate.yaml`, run the verifier:
+
+```bash
+agents-shipgate verify --base origin/main --head HEAD --json
+```
+
+Read `agents-shipgate-reports/agent-handoff.json` first and lead with
+`gate.merge_verdict` (a deterministic projection of `release_decision.decision`,
+which remains the gate in `report.json`), then the authoritative substrate
+`agents-shipgate-reports/verifier.json` and supporting/provisional
+`capability_review.top_changes[]`. Do not claim completion when
+`merge_verdict` is `blocked`, `insufficient_evidence`, or
+`human_review_required` unless the user accepted the human-review requirement, and
+never weaken `shipgate.yaml`, Shipgate CI, `AGENTS.md`, policies, baselines, or
+waivers to make Shipgate pass.