Agent Plugins Marketplace
← All plugins

architect-critic

v0.7.0

Anti-sycophancy reviewer for specs and plans, with optional cross-agent fresh-frame critique (synchronous or async background Codex audits).

CodexClaude Code6 Skills

By PrasLicense: MIT0 GitHub starsUpdated 2 weeks ago

Directory evidence

Runtimes
Codex and Claude Code
Parsed components
6 skill or MCP entries
Source updated
Oct 4, 2026
Manifest status
Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology →

Install architect-critic for Codex and Claude Code

Installs for the current user
codex plugin marketplace add pulseai-labs/claude-agent-scaffolding
codex plugin marketplace upgrade claude-agent-scaffolding-codex
codex plugin add architect-critic@claude-agent-scaffolding-codex

Paste and run these commands in a terminal with Codex. They add and refresh the claude-agent-scaffolding-codex catalog, then install this plugin.

Compatibility: the page URL and API slug “architect-critic” remain stable.

  • Codex: architect-critic@agent-plugin-marketplace → architect-critic@claude-agent-scaffolding-codex

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually
git clone https://github.com/pulseai-labs/claude-agent-scaffolding

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The plugin root is architect-critic/.

Plugin files

architect-critic/
├── .codex-plugin/plugin.json
├── .claude-plugin/plugin.json
├── skills/checking-adversary-readiness/SKILL.md
├── skills/critiquing-spec/SKILL.md
├── skills/listing-principles/SKILL.md
├── skills/managing-async-critique/SKILL.md
├── skills/promoting-principle/SKILL.md
└── skills/reviewing-critique-history/SKILL.md

Included Skills6

checking-adversary-readinessskills/checking-adversary-readiness/SKILL.md

Check whether the external adversary (Codex/Claude) is installed, authenticated, and schema-capable before a deep critique. Triggers on "check adversary readiness", "is codex ready", "critique doctor", "/critique-doctor", "can I run an async critique". Advisory and fail-soft — never blocks, never auto-installs.

critiquing-specskills/critiquing-spec/SKILL.md

Adversarial audit of a written spec or plan. Triggers on phrases like "audit this spec", "critique X", "adversarial review", "challenge the spec", "deep audit", "fresh-frame review". Runs host-agent self-audit in conversation; optionally invokes the other agent as a fresh-frame adversary at close-depth (--close flag or NL trigger): Codex when hosted in Claude Code, Claude Code when hosted in Codex. Produces challenges/gaps/alternatives, runs sequential rebuttal cycle with 1-5 concession scoring, appends run to state.json.

listing-principlesskills/listing-principles/SKILL.md

Show all principles in effect (shipped defaults + user-promoted + project-scoped + memory-bank patterns) grouped by source with annotations. Triggers on "show principles", "list principles", "what principles apply here", "principles in effect", "audit principles". Optional --source all|shipped|user|project filter.

managing-async-critiqueskills/managing-async-critique/SKILL.md

Manage background (async) close-depth critique jobs — status, result, cancel, and resume. Triggers on "critique jobs", "resume critique", "resume the async audit", "cancel critique audit", "check critique job", "/critique-jobs". Resume consolidates the finished Codex result with the persisted host self-audit and runs one unified rebuttal; a concluded run resumes inspect-only.

promoting-principleskills/promoting-principle/SKILL.md

Promote a principle to user-global or project-scoped principles.md. Triggers on "promote this principle", "record this as a principle", "add to principles.md", "make this a principle", "save this principle". Validates uniqueness, tags with source + timestamp, records the promotion in state.json.

reviewing-critique-historyskills/reviewing-critique-history/SKILL.md

Show recent architect-critic runs as a table, plus any in-flight background (async) audits. Triggers on "show recent critiques", "critique history", "list pending audits", "recent audits", "audit log", "in-flight critiques", "background audits". Reads state.json schema v3 (recent_runs[] + external_runs[]). Default N=10 rows; --limit N overrides.

Plugin manifests2

architect-critic/.codex-plugin/plugin.json
{
  "name": "architect-critic",
  "version": "0.7.0",
  "description": "Anti-sycophancy reviewer for specs and plans, with optional cross-agent fresh-frame critique (synchronous or async background Codex audits).",
  "author": {
    "name": "Pras"
  },
  "repository": "https://github.com/draco28/claude-agent-scaffolding",
  "license": "MIT",
  "keywords": [
    "critique",
    "architecture",
    "review",
    "planning"
  ],
  "skills": "./skills/",
  "interface": {
    "displayName": "Architect Critic",
    "shortDescription": "Adversarial review for specs, plans, and design choices.",
    "longDescription": "Architect Critic audits written specs and plans, surfaces challenges, runs a rebuttal cycle, records critique history, and can use a second agent as a fresh-frame adversary at close depth — synchronously, or as a managed background (async) job with status/result/cancel/resume and a readiness doctor.",
    "developerName": "Pras",
    "category": "Coding",
    "capabilities": [
      "Interactive",
      "Read",
      "Write"
    ],
    "defaultPrompt": [
      "Critique this spec at close depth.",
      "Critique this large spec at close depth in the background, then I'll resume.",
      "Check whether the external adversary is ready.",
      "Review recent critique history."
    ],
    "websiteURL": "https://github.com/draco28/claude-agent-scaffolding/tree/main/architect-critic",
    "brandColor": "#7C3AED"
  }
}
architect-critic/.claude-plugin/plugin.json
{
  "name": "architect-critic",
  "version": "0.7.0",
  "description": "Anti-sycophancy reviewer. 6 skills (gerund-named): critiquing-spec (audit + sequential rebuttal with T=4 concession scoring), managing-async-critique (background Codex audits: status/result/cancel/resume), checking-adversary-readiness (doctor), reviewing-critique-history, listing-principles, promoting-principle (manual). Ships ghost-notes + CORE shipped-default principles. Codex 0.125+ adversarial fresh-frame at close-depth, synchronous OR async (defer-to-resume unified rebuttal, durable state.json v3 job memory). Standalone-invocable; consumer plugins invoke skills in-conversation (no file IPC).",
  "author": {
    "name": "Pras"
  },
  "category": "workflow"
}

For maintainers

If you maintain this plugin, link to this source-backed listing from your README so users can review its manifest and indexed components.

[architect-critic on Agent Plugins Marketplace](https://pluginsmp.com/plugins/architect-critic)