--- title: Agent skills description: Share repo-specific agent workflows so MCP clients can discover and load your skill files alongside docs. --- Your documentation tells agents *what* your product does. Agent skills tell them *how* to work in your repository: review checklists, doc-writing pipelines, release steps, and other repeatable workflows you encode as markdown. ## Overview [Agent skills](https://cursor.com/docs/skills) are instruction files you commit to your repository. docs.page scans `.agents/skills/**` on the same Git ref as your hosted docs and exposes each `SKILL.md` as an MCP resource on your site's [MCP server](/features/mcp-server). There is no separate skills toggle. When MCP is enabled and your repo is public, skills appear automatically in the server's resource catalog alongside the `docs.json` schema resource. ## How it works ### Repository layout Each skill lives in its own folder under `.agents/skills/`. The entry point must be named `SKILL.md`: ```text .agents/skills/ docs-write/ SKILL.md release/ SKILL.md templates.md ``` Nested folders are supported. A file at `.agents/skills/team/deploy/SKILL.md` registers as skill slug `team/deploy`. Only paths that match `.agents/skills/**/SKILL.md` are published. Supporting files in the same folder (such as `templates.md` or `style-lint.md`) stay in Git for authors and agents that read the repo directly; they are not registered as separate MCP resources. ### MCP resource URIs When docs.page builds the MCP descriptor, each skill becomes a markdown resource with a stable URI: ```text docs-page://skills/{slug} ``` For example, `.agents/skills/docs-write/SKILL.md` maps to `docs-page://skills/docs-write`. Slashes in nested paths are preserved and URL-encoded in the URI. Clients list resources through the MCP session, then fetch the skill body as `text/markdown`, the same file content committed on GitHub at the resolved ref. ### Same endpoint as your docs Skills ride on the per-repository MCP endpoint. Production, branch previews, and custom domains use the same URL patterns as [MCP server](/features/mcp-server). Discovery runs when the server starts a session: docs.page reads the GitHub tree for the current ref, registers every matching skill, and serves file contents on demand. If MCP is disabled (`mcp.enabled: false`) or the repository is private, neither docs tools nor skill resources are available. ## Enable it 1. **Add skills to your repo.** Create one folder per skill under `.agents/skills/` and put instructions in `SKILL.md`. Use YAML frontmatter (`name`, `description`) when your client expects it. The body is plain markdown either way. 2. **Push to a public GitHub repository.** Skills follow the same visibility rules as hosted docs. Commit and push to the branch your MCP client targets (usually the default branch for production URLs). 3. **Connect an MCP client.** MCP is on by default. Open your live site, choose **Install with MCP** from the [page action menu](/features/public-github-hosting#page-actions) or **Install MCP** from **Cmd+K**, and add the endpoint to Cursor, Claude Code, or another HTTP MCP client. See [MCP server](/ai-agents/mcp-server) for endpoint URLs and client setup. After connecting, list MCP resources on your docs server. You should see one entry per `SKILL.md`, plus `docs-page-config-schema`. Fetch a skill resource to load its instructions into the agent context. ## Related Per-repo endpoint, tools, and how to connect a client. Text exports for LLM discovery without MCP. Why docs.page reads from public repositories at request time. URL patterns for MCP and other published routes.