Skills

anywhere-agents ships four skills. Each is a self-contained capability with its own SKILL.md definition. The router (my-router) dispatches automatically based on prompt keywords, file types, and directory hints.

•   implement-review

  ---

  Structured dual-agent review loop with content-type lenses. Codex reviews, Claude applies fixes, iterates until clean.

  Deep docs

•   my-router

  ---

  Context-aware dispatcher. Picks the right skill from prompt keywords, file types, and directory hints.

  Deep docs

•   ci-mockup-figure

  ---

  HTML mockups of systems, dashboards, timelines → space-efficient PNG / PDF figures via headless Chrome.

  Deep docs

•   readme-polish

  ---

  Audit and rewrite GitHub READMEs using modern 2025-2026 patterns (badges, hero, callouts, collapsibles).

  Deep docs

Skill lookup order

When a skill is invoked, the agent looks for its SKILL.md in this order:

1. Project-local: skills/<name>/SKILL.md in the consuming project. Use this to override a shared skill or add project-specific skills.
2. Pack-deployed: .claude/skills/<name>/SKILL.md installed by anywhere-agents pack install. The .claude/ prefix is a historical Claude Code convention; the SKILL.md contents are agent-agnostic.
3. Bootstrapped shared: .agent-config/repo/skills/<name>/SKILL.md in the consuming project (synced by bootstrap).
4. Plugin-provided: agent-platform plugins may provide skills too (e.g., Claude Code plugins).

The router follows the same lookup order.

Adding your own skill

In a fork of anywhere-agents:

1. Create skills/<your-skill>/SKILL.md with the skill definition.
2. Add agents/openai.yaml for Codex invocation and .claude/commands/<your-skill>.md for Claude Code invocation.
3. Register the skill in skills/my-router/references/routing-table.md with triggering keywords, file types, and directory hints.
4. Add regression tests where appropriate under tests/.

See the four shipped skills for reference structure.