Official agent skills and plugin definitions for working with Terminal, the unified API for fleet and telematics data.
The repository contains one installable terminal plugin and three portable Agent Skills:
terminal-fleet-data— answer operational questions about fleet data through Terminal MCP or the Terminal CLI.terminal-cli— install, configure, and troubleshoot the Terminal CLI.build-with-terminal— build product integrations using Terminal APIs, Link, and webhooks.
Install every skill into a supported agent:
npx skills add terminal-api/skillsOr install one skill:
npx skills add terminal-api/skills --skill terminal-fleet-dataThe installer supports Claude Code, Codex, Cursor, GitHub Copilot, OpenCode, and other Agent Skills clients.
The complete plugin adds the skills and configures Terminal's OAuth-enabled remote MCP server at https://mcp.withterminal.com/mcp.
/plugin marketplace add terminal-api/skills
/plugin install terminal@terminal
Workspace administrators can import https://github.com/terminal-api/skills from Workspace settings → Plugins → Import marketplace. Codex users can also install the standalone skills with npx skills add.
The first release declares the MCP server directly, so OpenAI surfaces may label the full plugin Desktop only until Terminal is also available as an approved connected app.
Until the plugin is listed in the public Cursor Marketplace, install the skills with npx skills add or import this repository into a Cursor team marketplace.
The portable plugin is in plugins/terminal. It follows the Agent Plugins 1.0 specification.
- MCP: Your client opens Terminal's OAuth flow. Access matches your Terminal Dashboard user and active organization.
- CLI: Install
@terminal-api/cliand configure a Terminal secret key locally. Never paste a secret key or connection token into a chat.
See the Terminal agents guide for current setup details.
npm ci
npm run checkThe plugin has its own version. It is intentionally independent from Terminal CLI and MCP server release versions.
Contributions are welcome. See CONTRIBUTING.md for authoring, validation, and versioning guidance.
Skills are executable instructions. Review their contents before installation and report concerns according to SECURITY.md.
MIT