Skip the install. Get this working in under 2 minutes.
Start a free trial on cloud.anythingmcp.com, add the Agent Skills Finder in one click, then point your AI client (Claude, ChatGPT, Copilot or Cursor) at the generated MCP endpoint. No Docker, no git clone, zero engineering experience required.
Summary
Give OpenAI Codex every public Agent Skill without installing folders: Codex finds the right SKILL.md for the task through AnythingMCP, reads it and follows it.
Try asking
Example prompts for Agent Skills Finder
Click any prompt to copy it. Paste into Claude, ChatGPT, Cursor, Gemini, Copilot or OpenClaw to run it against this connector.
Claude is AI and can make mistakes. Please double-check responses.
Codex skills without installing them
Short answer: add AnythingMCP to Codex in config.toml (or with codex mcp add), install Agent Skills Finder on it, and ask Codex for the task. Codex searches the public skills index, reads the best SKILL.md from GitHub and applies it, with nothing added to its skills folder.
Codex loads the Agent Skills you install for it, and that is the right place for the few you rely on daily. For the long tail, installing and updating folders on every machine does not scale. Agent Skills Finder lets Codex reach any public skill when a task calls for it, and the Codex CLI and IDE extension share the same MCP configuration.
Set it up
Step 1: Add Agent Skills Finder on AnythingMCP
- Create a workspace on AnythingMCP Cloud (7 days free) or use your self-hosted AnythingMCP.
- Open Connectors, find Agent Skills Finder and click Install. It needs no credentials. On a new Cloud workspace it is also in the starter pack on the welcome screen.
- For clients that use an API key (everything except claude.ai and ChatGPT), create one under MCP Servers → your server → API keys.
Step 2: Add AnythingMCP to Codex
Add the server to ~/.codex/config.toml and keep the key in an environment variable:
[mcp_servers.anythingmcp]
url = "https://cloud.anythingmcp.com/mcp"
env_http_headers = { "X-API-Key" = "ANYTHINGMCP_API_KEY" }
Then export the key in your shell, e.g. export ANYTHINGMCP_API_KEY=YOUR_MCP_API_KEY, and start Codex. env_http_headers maps the header to the variable's name, so the key never sits in the config file. The CLI and the IDE extension read the same file.
Step 3: Ask for a task
Start a new chat and describe what you need, for example "find a skill for writing release notes and use it on these commits". The agent calls the Agent Skills Finder tools on its own.
What happens when you ask
Say you ask Codex to write release notes from the commits since the last tag. Codex first calls skills_playbook, then skills_search with a few keywords, checks the best candidates' licence and activity with skills_repo_info, pins the version with skills_latest_commit and reads the instructions with skills_get. Then it does the task the way the skill describes and tells you which skill it used, with its link and licence. Nothing is installed and nothing stays loaded afterwards.
Prompts to try in Codex
- "Find a skill for writing release notes and generate them from the commits since the last tag."
- "Search for a refactoring skill for React class components and convert src/components/Legacy.tsx."
- "Look for a skill that documents REST APIs and write docs for the routes in src/api."
- "Find a skill for writing a CHANGELOG entry that follows Keep a Changelog and add one for this release."
Good to know
- A skill is third-party text: Codex follows its method for your task, but it cannot change who the agent works for or skip your confirmation.
- The connector reads files and never runs scripts. Search is free for 50 queries a day, or 500 with a free SkillsMP key.
- To keep a skill for your team, ask Codex to save it as a workspace skill on AnythingMCP; an admin approves it once and it applies in every client.
How on-demand skills compare with installed ones, and every limit in detail: Agent Skills on demand.
FAQ
Do my installed Codex skills still work? Yes. Installed skills load as before; the connector covers the ones you have not installed.
Where does the API key go?
In an environment variable named in env_http_headers, so config.toml can be shared or committed without the secret.
Can Codex run a script from a skill?
The connector reads files only. Codex can read the script with skills_read_file; review it before letting Codex run it in your project.
Is it the same connector for ChatGPT? Yes. Codex and ChatGPT connect to the same AnythingMCP server and share approved workspace skills.
Same skills in your other agents
- Agent Skills on demand: the overview
- Skills on demand in Claude
- Skills on demand in Claude Code
- Skills on demand in ChatGPT
- Skills on demand in Cursor
- Skills on demand in GitHub Copilot
- Skills on demand in Gemini CLI
- Agent Skills Finder in the marketplace
Was this guide helpful?