AI agents: fetch the documentation index at llms.txt. Markdown versions are available by appending .md to any page URL, including this page's markdown.
Skills
Teach AI coding assistants how to use the Done Bear CLI with installable agent skills.
Done Bear skills are instruction bundles that teach AI coding assistants how to operate the Done Bear CLI. Install a skill and your agent gains command discovery, auth handling, task workflows, and safety-aware execution. No custom prompts needed.
Skills work with Claude Code, OpenCode, Codex, and Cursor.
Before you begin
- Install Node.js 18 or newer.
- Install the Done Bear CLI globally:
npm install -g donebear. - Sign in with
donebear auth login.
Install
npx skills add donebear/skills -g --all -yTo verify, open your AI coding assistant and ask it to run a Done Bear command:
List my open tasks in Done Bear
The assistant should run donebear task list --json and return your tasks. If the CLI isn't authenticated, the skill guides the assistant through donebear auth login automatically.
What a skill does
A skill gives your AI assistant:
- Command knowledge: common commands and flags without running
donebear specfirst. - Safety tiers: read-only commands run freely, destructive commands require confirmation.
- Error recovery: auth failures, missing workspaces, and unknown flags are handled automatically.
- JSON-first output: agents parse
--jsonoutput instead of scraping text.
How skills differ from MCP
The MCP server gives agents structured remote tool calls. Skills give agents CLI fluency. They run donebear commands directly in the terminal.
Use MCP when your client supports it and you want the cleanest integration. Use skills when you want agents to run the CLI directly, chain shell commands, or work in environments where MCP isn't available.
How skills differ from the CLI
You still need the Done Bear CLI installed globally (npm install -g donebear). The skill teaches your agent how to use it, so you don't need to learn the commands yourself.
Next steps
- Read MCP overview if your client supports the Model Context Protocol.
- Read CLI overview for the full command surface.