agent-skills

Small, composable workflow skills for coding agents: docs-first discipline, autonomous execution, cross-agent watchdog and plan arbitration, efficient model orchestration, status recaps, usage-limit awareness, and local visual plan/recap modes.

Skills

  • agent-watchdog

    user-invokable

    Watch, audit, or fix another agent's work from outside its context.

  • efficient-fable

    user-invokable

    Orchestrate codebase-heavy work efficiently when running as Claude Fable.

  • efficient-frontier

    user-invokable

    Run a high-cost frontier model efficiently on codebase-heavy work.

  • feedback

    user-invokable

    Save and apply your agent-skills workflow preferences.

  • plan-arbiter

    user-invokable

    Compare, merge, and arbitrate competing plans.

  • plow-ahead

    user-invokable

    Make autonomous progress without routine clarification stops.

  • quick-recap

    user-invokable

    End each response with a clear status signal.

  • read-the-damn-docs

    user-invokable

    Pull and read real documentation before implementing or integrating.

  • stay-within-limits

    user-invokable

    Check usage limits before and between heavy agent work.

  • visual-plan

    user-invokable

    Turn a plan into a self-contained interactive HTML document.

  • visual-recap

    user-invokable

    Turn a PR, branch, or commit diff into an interactive HTML recap.

Keywords

Install

/plugin marketplace add joshuaoliphant/claude-plugins
/plugin install agent-skills@oliphant-plugins

← All plugins