Jido Skills

Builder-oriented Codex skills for the Jido ecosystem.

This repository contains a generated catalog of skills for the public Jido ecosystem, plus a router skill that helps choose the right package skill for a task. The current catalog was generated automatically by ChatGPT Codex from public Jido ecosystem material, including Jido docs, HexDocs, hex.pm, and related source references, then validated with the local skill-creator tooling.

What Is Included

• 12 package skills for the current Jido ecosystem:
  • llm-db
  • req-llm
  • jido-action
  • jido-signal
  • jido
  • jido-browser
  • jido-memory
  • jido-behaviortree
  • ash-jido
  • jido-studio
  • jido-messaging
  • jido-otel
• 1 router skill:
  • jido-skill-router
• 1 generation prompt:
  • source/prompts.md

How To Use

If you already know the package boundary, invoke the package skill directly.

Examples:

Use $jido-action to scaffold a new Jido.Action for sending a webhook.
Use $req-llm to add streaming support for Anthropic responses.
Use $ash-jido to expose these Ash actions to a Jido agent.

If you do not know which package skill to start with, use the router skill first.

Use $jido-skill-router to choose the right Jido ecosystem skills for an Ash-triggered agent workflow.
Use $jido-skill-router to route this task across Jido memory, signals, and OpenTelemetry.

The router skill will:

• identify the anchor package skill
• pull in adjacent skills only when the task crosses their boundaries
• avoid loading the entire ecosystem by default

Repository Layout

• skills: all generated skills
• skills/jido-skill-router: router skill for package selection and multi-skill handoffs
• skills/jido-skill-router/references/skill-manifest.yaml: machine-readable manifest for the current skill catalog
• source/prompts.md: generation prompt used to create the package skills

Notes

• This catalog was generated automatically by ChatGPT Codex.
• It is grounded in public Jido docs, but some ecosystem packages currently have thinner public documentation than others.
• When docs are thin, the corresponding skills are intentionally narrow instead of guessing unsupported behavior.
• Treat this repository as a practical starting point and review package-specific guidance before treating it as authoritative.

Regenerating

To regenerate or extend the catalog:

1. Update source/prompts.md.
2. Run the prompt through Codex or another compatible agent.
3. Validate generated SKILL.md files with the local skill-creator validator.