An agentic skill for Claude Code and Cowork, packaged as an installable plugin marketplace:
| Plugin | What it does |
|---|---|
| cto-council | A multi-persona code and spec review council. 15 specialized reviewers (architecture, security, performance, LLM-security, compliance, accessibility, API-contract, code-quality, senior-engineer, spec-compliance, a coverage meta-pass, and four stack reviewers for Java / Rails / Vue / C++-DB2) run as parallel subagents. An Engineering Manager orchestrates them, deduplicates findings, and delivers one verdict with an audit-score and top-3 action items. |
In Claude Code (or Cowork), add this marketplace and install the plugin:
/plugin marketplace add shaase-ctrl/agentic-skills
/plugin install cto-council@agentic-skills
cto-council - review code, a PR, a spec, or a whole repo:
/cto-council:council-review
or just say "council this branch", "review this PRD against my code in ~/projects/foo",
"security review this". The Engineering Manager picks the right reviewer sequence for the asset.
- cto-council runs each reviewer as its own subagent with its own context window and persona briefing, with built-in defenses against known LLM-reviewer failure modes (MAST taxonomy, OWASP LLM Top 10).
- This repo is generated. Do not edit it directly; changes are made in the private source registry and published from there, so a fix cannot land in one place and be lost in the other.
research-engine was published here until 2026-08-07 and has been temporarily withdrawn while
it is restructured. It will return once its publishable core is cleanly separated from
material that is not mine to share.
MIT, see LICENSE. Author: Sascha Haase.