Using with Other Agents

Skills on inference shell follow the Agent Skills open standard. This means skills you create here work with Claude Code, Cursor, Windsurf, and any tool that supports the spec.

The Skill URL

Every skill has a stable content URL:

1https://api.inference.sh/skills/{namespace}/{name}/content

This URL serves the skill's SKILL.md content (frontmatter + instructions). It can be used directly in any agent config that accepts a URL for context or instructions.

Supporting files are available at:

1https://api.inference.sh/skills/{namespace}/{name}/files/{path}

For example: .../files/references/api-docs.md, .../files/scripts/helper.py

Claude Code

Claude Code discovers skills from the local filesystem. To use an inference shell skill:

Manual Install

Download the skill files and place them in Claude Code's skills directory:

1~/.claude/skills/{skill-name}/2├── SKILL.md3├── references/4│   └── api-docs.md5└── scripts/6    └── helper.py

The download endpoint provides a manifest of all files with their URLs:

1GET https://api.inference.sh/skills/{namespace}/{name}/download

Returns:

1{2  "name": "code-review",3  "namespace": "inferencesh",4  "version": "1.0.0",5  "files": [6    { "path": "SKILL.md", "uri": "https://cdn.inference.sh/..." },7    { "path": "references/checklist.md", "uri": "https://cdn.inference.sh/..." }8  ]9}

Download each file from its uri and write to the corresponding path under ~/.claude/skills/{name}/.

Belt CLI

1belt skill use inferencesh/code-review

Prints the skill's main instructions (SKILL.md body) to stdout. Public skills do not require authentication.

When a skill has supplementary files (references/, scripts/, examples/), the CLI prints tips after the instructions:

1  tip: this skill has 3 supplementary file(s) not included above.2    install locally:  belt skill add inferencesh/code-review3    list files:       belt skill files inferencesh/code-review

• belt skill add — downloads all files into your local skills directory (for Claude Code, Cursor, etc.)
• belt skill files — lists supporting file paths without downloading

Belt plugin init (all supported agents)

To install belt skills and lifecycle hooks in one step:

1belt login2belt plugin init claude    # Claude Code (plugin marketplace)3belt plugin init codex     # OpenAI Codex (requires Codex CLI)4belt plugin init cursor    # Cursor5belt plugin init gemini    # Gemini CLI6belt plugin init opencode  # OpenCode7belt plugin init pi        # Pi8belt plugin init windsurf  # Windsurf

Claude Code and Codex use their respective plugin marketplaces (belt-sh/skills). That installs the full belt plugin (hooks, skills, bin scripts, agents, rules). Codex requires the Codex CLI on your PATH.

Cursor, Gemini CLI, OpenCode, Pi, and Windsurf use a manual install: five core skills (suggest, skill, apps, remember, belt) plus a hooks config that calls belt suggest --json on prompt submit and belt review --agent <name> when a session ends. Cursor also runs a session-start hook that prints belt me output or a login reminder (manual install) or, with the marketplace plugin, runs the full device authorization flow when you are not logged in. Cursor includes a PreCompact hook that forces review before context compaction. OpenCode installs a TypeScript plugin under ~/.config/opencode/plugins/belt/ (npm install + opencode plugin --global when the OpenCode CLI is on PATH). Pi installs a TypeScript extension at ~/.pi/agent/extensions/belt.ts (pi install when the Pi CLI is on PATH). Skills are pulled from the platform when you are logged in; embedded copies are used when you are not logged in, offline, or the download fails.

See Coding agents for Claude Code setup. Codex plugin details: Codex plugin.

After belt plugin init codex, run /hooks in Codex to review and trust belt hooks. If you previously had belt hooks in ~/.codex/hooks.json from a manual or fallback install, plugin init removes those entries automatically so hooks do not run twice. Restart Cursor, Gemini CLI, OpenCode, Pi, or Windsurf after init.

Cursor, Windsurf, Codex, and Gemini CLI

Use belt plugin init <agent> to install belt skills and lifecycle hooks:

1belt plugin init cursor2belt plugin init windsurf3belt plugin init codex4belt plugin init gemini5belt plugin init opencode6belt plugin init pi

For Codex, belt plugin init codex uses the Codex plugin marketplace when the codex CLI is available (full plugin install). The Codex CLI must be on your PATH.

For Cursor, Gemini CLI, OpenCode, Pi, and Windsurf, each command installs five core belt skills (suggest, skill, apps, remember, belt) into the agent's skills directory and copies a hooks config. Skills are downloaded from the platform when you are logged in; embedded copies are used when you are not logged in, offline, or the download fails. Hooks run belt suggest --json on prompt submit and belt review --agent <name> on session end to capture reusable knowledge. Cursor's session-start hook prints belt me output when logged in or a login reminder when not (manual install). Claude Code and Codex plugin installs use belt plugin hook session-start, which runs device authorization automatically when no API key is configured.

See Coding agents for full setup details.

Manual install

Any agent that supports the Agent Skills standard can use inference shell skills without belt plugin init:

1. Get the skill files via the download endpoint or skill URL
2. Place them where the agent expects to find skills
3. The agent discovers and uses them like any other skill

Check your agent's documentation for where skills are stored and how they're discovered.

URL-Based Usage

Some agents support loading skills directly from a URL rather than the filesystem. In that case, point to the content URL:

1https://api.inference.sh/skills/{namespace}/{name}/content

This returns the full SKILL.md content — frontmatter + instructions — ready for the agent to parse.

The Roundtrip Guarantee

Skills created on inference shell produce the exact same SKILL.md format as the open standard. A skill created in the web editor and installed locally is indistinguishable from one written by hand:

• All spec frontmatter fields: name, description, license, compatibility, metadata, allowed-tools
• Same markdown body
• Same directory structure (references/, scripts/, examples/, templates/)

This is by design — inference shell is a creation and hosting platform, not a proprietary format.

Next

• Creating skills — build skills on inference shell
• The registry — browse and share public skills
• Agent Skills standard — the open spec