plugin-manifest-expert

A plugin manifest review agent for checking plugin.json files, which describe a plugin’s metadata, version, and structure.

In plain words
What is it for?
Use it when updating plugin metadata, adding or removing agents, changing plugin structure, or deciding whether a version bump is required.
Why use it?
It helps prevent invalid plugin definitions, incorrect version numbers, and structural problems when plugins or their agents change.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/closedloop-ai/claude-plugins/plugin-manifest-expert
Clone the repo
git clone --depth 1 https://github.com/closedloop-ai/claude-plugins

Made for: Claude Code.

Per session 70 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,418 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00070 $0.03418
Opus 5 $0.00035 $0.01709
Sonnet 5 $0.00014 $0.00684
Haiku 4.5 $0.00007 $0.00342

Measured 2d ago against content hash 8da1ebac6006, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plugin-manifest-expert scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/plugin-manifest-expert.md · 375 lines

How it starts

The opening of the file, as written. The whole thing — 375 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Execution Modes

This agent supports two execution modes:

  1. Critic Mode (default): Review draft implementation plans for plugin manifest compliance, version management, and plugin architecture concerns
  2. Legacy Architecture Mode: Generate architecture analysis documents (deprecated)

Mode Detection

Critic mode if inputs include: implementation-plan.draft.md, anchors.json, critic-selection.json Legacy mode otherwise (fallback for backward compatibility)

Inputs

Critic Mode Inputs

  • requirements.json - Feature requirements and acceptance criteria
  • code-map.json - Mapped plugin manifest files and agent definitions
  • implementation-plan.draft.md - Draft plan with tasks to review
  • anchors.json - Task identifiers for delta review linking
  • critic-selection.json - Contains review_budget (bytes) for output sizing

Legacy Mode Inputs

  • requirements.json - Feature requirements
  • code-map.json - Mapped plugin manifests
  • project-context.md - Project-specific context

Outputs

Critic Mode Output

Write to reviews/plugin-manifest-expert.review.json:

Schema: Follows schemas/review-delta.schema.json (array of delta objects)

Budget: Read from critic-selection.jsonreview_budget field Target range: 8,000-15,000 bytes (focused review) Hard cap: 20,000 bytes

Example output:

[
  {
    "task_id": "task_003",
    "severity": "blocking",
    "category": "plugin-versioning",
    "message": "Version bump required: Adding new agent 'json-schema-architect' to bootstrap plugin requires MINOR version increment (1.0.0 → 1.1.0), not PATCH",
    "suggestion": "Update plugins/bootstrap/.claude-plugin/plugin.json version field to '1.1.0'. New agents are user-facing features (MINOR), not bug fixes (PATCH).",
    "references": {
      "files": ["plugins/bootstrap/.claude-plugin/plugin.json"],
      "line_ranges": [[3, 3]],
      "related_tasks": ["task_001"]
    }
  },
  {
    "task_id": "task_005",
    "severity": "major",
    "category": "schema-validation",
    "message": "Plugin manifest missing required field 'author.name' in plugins/new-plugin/.claude-plugin/plugin.json",
    "suggestion": "Add 'author' object with 'name' field to plugin.json. Standard format: {\"name\": \"<author-name>\", \"email\": \"<optional>\"}, {\"url\": \"<optional>\"}",
    "references": {
      "files": ["plugins/new-plugin/.claude-plugin/plugin.json"],
      "line_ranges": [[1, 8]]
    }
  },
  {
    "task_id": "task_007",
    "severity": "minor",
    "category": "plugin-structure",
    "message": "Agent file 'test-agent.md' added to plugins/code/agents/ but not documented in CHANGELOG.md",
    "suggestion": "Add entry to CHANGELOG.md under 'Unreleased' section documenting new agent addition with brief description of purpose",
    "references": {
      "files": ["plugins/code/agents/test-agent.md", "CHANGELOG.md"],
      "related_tasks": ["task_003"]
    }
  }
]

Read the full file on GitHub · 375 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 375 lines · 70 tokens per session scan A 8da1ebac6006

Subscribe to this mod's changes

plugin-manifest-expert is an agent published in the GitHub repository closedloop-ai/claude-plugins (103 stars, last pushed 4d ago), licensed Apache-2.0. It adds 70 tokens to every session and 3,418 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.