agentjido/jido_run/jido

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Builder-oriented guidance for the upstream `jido` package. Use when Codex needs to design or review Jido agents, directives, action pipelines, runtime loops, or example applications that compose `jido_action`, `jido_signal`, memory, and AI-facing packages into working Elixir systems.

Install

npx vskill@latest install agentjido/jido_run/jido

SKILL.md

Jido

jido is the upstream Hex package name.

Start Here

Use this skill for the core agent runtime and orchestration layer.

Good triggers:

  • "Design a Jido agent for this workflow."
  • "Refactor this runtime loop into actions, directives, and signals."
  • "Turn the Jido docs into a runnable agent example."
  • "Review whether a feature belongs in core Jido or a package-specific extension."

Read references/builder-notes.md before coding when the task spans multiple Jido concepts or packages.

Primary Workflows

Design an agent runtime
  • Start with the agent boundary: state, directives, allowed actions, emitted signals, and external dependencies.
  • Keep planning and execution explicit. Do not smuggle hidden orchestration into one action or one callback.
  • Prefer small, composable workflows that can be tested without a full application shell.
Compose ecosystem packages
  • Reach for jido-action when the job is an executable unit of work.
  • Reach for jido-signal when the job is event shape and dispatch.
  • Reach for jido-memory, req-llm, llm-db, or ash-jido only when the feature truly crosses into those domains.
Turn docs into runnable examples
  • Build a thin end-to-end example: agent definition, one or two actions, one signal path, and one observable result.
  • Prefer examples that demonstrate runtime decisions or directives over static data transforms.
  • Keep examples executable in tests or iex, not locked to a large demo app.
Review boundaries
  • Keep core runtime coordination in jido.
  • Push provider, catalog, browser, memory, and Ash-specific details into their package skills.
  • Narrow proposals that require undocumented runtime hooks.

Build Checklist

  • Define the agent state and lifecycle before writing callbacks.
  • List the actions, directives, and signals that participate.
  • Identify external integrations and keep them behind package boundaries.
  • Add tests for state transitions, directive behavior, and failure recovery.

Boundaries

  • Do not use this skill for package-specific transport or provider code when another package already owns that boundary.
  • Do not invent runtime primitives that the docs do not support.
  • Do not let one agent absorb unrelated business workflows that should remain separate.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
44
Repository forks
20
7-day trend score
19

Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 17 May 2026, 5 months ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/agentjido/jido_run/jido/badge)](https://verified-skill.com/skills/agentjido/jido_run/jido)

Something wrong with this skill or its page? Tell us.