Agent Management
Wraps the ai-devkit CLI to list, start, assign, and stop multi-agent workers.
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 20, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 55a5bfb
Fetched the repo tree (23 SKILL.md files in a plugin monorepo) and read skills/agent-management/SKILL.md raw plus packages/cli/package.json — the skill is a clean wrapper around the real published `ai-devkit` npm CLI (v0.47.0, MIT) backed by @ai-devkit/agent-manager. No exfiltration, base64, secrets, or hidden instructions in the SKILL.md; only plain documented subcommands. INSTALL 4/5 (frontmatter valid with name+description, the one referenced dep `$agent-communication` exists at skills/agent-communication/SKILL.md, clean install path — docked 1 because copying files is inert without the external CLI and running worker agents). TRIGGER 5/5, phrasings that SHOULD load: 1) "List the running AI DevKit agents and tell me which one I am" 2) "Start a codex worker to review the auth middleware and assign it that task" 3) "Stop the idle agent named ui-tests and resume my earlier session"; should NOT: 4) "Refactor this function to be more readable" 5) "What's this repo's git commit message convention?". OUTPUT NOT measured (not a measured loss): requires the external ai-devkit CLI and live spawned agents I will not install/run, so no baseline-vs-skill comparison is possible. DOCS 4/5 — description matches implementation and CLI docs exist; no overselling.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 4/5
- Triggers reliably 5/5
- Output vs. baseline 5/10
- Docs & honesty 4/5
What Agent Management does
A thin prompt wrapper that teaches the agent to orchestrate other AI worker agents via `ai-devkit agent ...` subcommands (list, detail, start, send, sessions, rename, kill). It triggers when you need to identify yourself among running agents, spin up a codex worker, delegate scoped work, or tear an agent down. Pure prompt guidance — it does nothing without the external ai-devkit CLI and live worker agents installed.
How to install Agent Management
Copy skills/agent-management/ into ~/.claude/skills/ (or project .claude/skills/), then install the ai-devkit CLI globally (npm i -g ai-devkit) or invoke via npx ai-devkit@latest. Worker agents (e.g. codex) must be running for commands to do anything.
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Agent Management
-
/agent-managementWraps the ai-devkit CLI to list, start, assign, and stop multi-agent workers.
It also activates on plain-language prompts like these:
-
List all currently running AI agents now -
Start a new worker agent for this task -
Delegate this work to another agent
Frequently asked questions
- Is the Agent Management skill free?
- Yes. The skill itself is free from codeaholicguy/ai-devkit. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Agent Management work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 20, 2026. Verdict: Works with setup. Fetched the repo tree (23 SKILL.md files in a plugin monorepo) and read skills/agent-management/SKILL.md raw plus packages/cli/package.json — the skill is a clean wrapper around the real published `ai-devkit` npm CLI (v0.47.0, MIT) backed by @ai-devkit/agent-manager. No exfiltration, base64, secrets, or hidden instructions in the SKILL.md; only plain documented subcommands. INSTALL 4/5 (frontmatter valid with name+description, the one referenced dep `$agent-communication` exists at skills/agent-communication/SKILL.md, clean install path — docked 1 because copying files is inert without the external CLI and running worker agents). TRIGGER 5/5, phrasings that SHOULD load: 1) "List the running AI DevKit agents and tell me which one I am" 2) "Start a codex worker to review the auth middleware and assign it that task" 3) "Stop the idle agent named ui-tests and resume my earlier session"; should NOT: 4) "Refactor this function to be more readable" 5) "What's this repo's git commit message convention?". OUTPUT NOT measured (not a measured loss): requires the external ai-devkit CLI and live spawned agents I will not install/run, so no baseline-vs-skill comparison is possible. DOCS 4/5 — description matches implementation and CLI docs exist; no overselling.
- What is the Agent Management SkillProof Score?
- 7.2/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.
- How do I install Agent Management?
- 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 Agent Management 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.