Agent skills by riteofstring. Each skill is a top-level folder with a
SKILL.md, and works in Claude Code, Codex and other agents that read the
same format.
- overseer: supervised implementation. Two reviewers must agree on the plan before any code, a delegated coder builds it, the coordinator checks the work against evidence, a read-only auditor digs through the code, and fresh scope and direction checks run at each checkpoint. If you ask for polish or a target score up front, it hands the verified build to council.
- council: a council of fresh, independent
reviewer agents score the work out of 10 from chosen perspectives (design,
UX, marketing, performance and so on). The main agent fixes what they find,
and new reviewers score it again until every reviewer reaches the target
score, or until the feedback stops being useful. Has a
polishmode for detail-oriented finish.
npx skills add riteofstring/skillsOr copy or symlink a skill folder into ~/.claude/skills/ (Claude Code) or
~/.agents/skills/ (Codex and others).
/overseer add per-machine sorting and pinning
/overseer add a credentials page, then polish design and UX to 9 with council
/council design and UX for the settings screens, target 9
/council polish the landing page
The second line builds under overseer first. Once the build is verified, it reports back and runs council on the result with a target of 9.