Sub-Agents

Spawns external CLI AIs (codex, cursor-agent, gemini, grok...) as scoped sub-agents

Tested · Works

Test report

Verdict
Tested · Works
Score
9.2/10
Tested
Jul 14, 2026
Environment
Claude Code 2.x (agent harness)
Upstream re-checked
Jul 18, 2026 · 89dba38

Every claim checked out: the runner script and all 5 sibling _builder/_constants/_executor/_loader/_resolver modules it imports are real and present, and following the SKILL.md turned a vague 'delegate this to an agent' request into an exact, runnable script invocation with the correct flags and JSON-status handling. Main watch-out is the description's generic 'delegates a task to an agent' phrasing risks colliding with a host's own native sub-agent/Task tooling rather than this skill's external-CLI use case.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

  • Installs cleanly 5/5
  • Triggers reliably 4/5
  • Output vs. baseline 9/10
  • Docs & honesty 5/5

What Sub-Agents does

Runs named agent definitions from a .agents/ directory through external CLI backends (codex, claude, cursor-agent, glm, grok, gemini, opencode) via a bundled Python runner script, returning structured JSON with status/exit_code. Triggers when a user names a specific agent or sub-agent to run, references an agent-definition file, or asks to delegate a task to one of these external CLI tools.

How to install Sub-Agents

git clone https://github.com/shinpr/sub-agents-skills
cd sub-agents-skills
mkdir -p ~/.claude/skills
cp -r plugins/runner/skills/sub-agents ~/.claude/skills/sub-agents

Skills live in ~/.claude/skills/ (global) or .claude/skills/ (per-project). Restart Claude Code after installing.

Commands — how to trigger Sub-Agents

  • /sub-agents Spawns external CLI AIs (codex, cursor-agent, gemini, grok...) as scoped sub-agents

It also activates on plain-language prompts like these:

  • Run the code-reviewer agent defined in .agents/ on this file
  • Delegate this refactor to the security-auditor sub-agent
  • Use codex as a sub-agent to review src/ and report back

Frequently asked questions

Is the Sub-Agents skill free?
Yes. The skill itself is free from shinpr/sub-agents-skills. SkillProof publishes the install command and an independent test verdict at no cost.
Does Sub-Agents work with Claude Code?
We tested it with Claude Code 2.x (agent harness) on Jul 14, 2026. Verdict: Tested · Works. Every claim checked out: the runner script and all 5 sibling _builder/_constants/_executor/_loader/_resolver modules it imports are real and present, and following the SKILL.md turned a vague 'delegate this to an agent' request into an exact, runnable script invocation with the correct flags and JSON-status handling. Main watch-out is the description's generic 'delegates a task to an agent' phrasing risks colliding with a host's own native sub-agent/Task tooling rather than this skill's external-CLI use case.
What is the Sub-Agents SkillProof Score?
9.2/10 — installs cleanly 5/5, triggers reliably 4/5, output vs. baseline 9/10, docs & honesty 5/5.
How do I install Sub-Agents?
Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use Sub-Agents with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.