# Aback Skills > A free, verified marketplace for agent skills. Authors publish a `SKILL.md` > for free. An admin reviews it. Once approved it is publicly available and > any agent that speaks the Model Context Protocol can install it into a > project with a single tool call. Publishing is free on every plan, with no listing fee, no upload limit and no revenue share. The paid plans pay for MCP access at scale, and nothing else. ## What a skill is A skill is a markdown file, `SKILL.md`, that teaches an agent how to do one specific thing well. It has sections for Instructions, Rules and Examples. It is a prompt, a procedure and a set of examples, in a format an agent can read at the point of use. ## How to use the catalogue The delivery channel is an MCP server at: POST https://abackskills.com/mcp It speaks the Model Context Protocol over Streamable HTTP and is stateless. Authenticate with an API key: Authorization: Bearer abk_live_ Create a key at https://abackskills.com/dashboard/mcp. The secret is shown exactly once and is never stored. Clients that cannot set headers may pass the key as an `api_key` query parameter instead. ## Tools | Tool | Purpose | | --- | --- | | `search_skills` | Search the approved catalogue by text, category or tag | | `get_skill` | Read one approved skill, with its risk signals and full content | | `install_skill` | Install a skill into a project and return the files to write | | `list_installed_skills` | List what is installed in a project | | `remove_skill` | Remove an installation from a project | | `list_projects` | List the install targets on the account | | `check_updates` | Compare installed versions against the latest approved version | | `get_usage` | Report the current allowance and how much of it is used | ## Installing a skill `install_skill` returns the exact file path and the exact file content. The server cannot write to your machine, so the client writes the file. The default target directory is `.agents/skills`, which is the cross-agent convention. Pass `target_dir` to use a different layout. { "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "install_skill", "arguments": { "slug": "code-review", "project": "my-app" } } } The result carries a `contentHash`. Verify the write against it. ## Reading a skill without installing it The content of an approved skill is also available as an MCP resource: aback://skills/{slug} aback://skills/{slug}/{version} And through the public REST API, which needs no key and has no quota: GET https://abackskills.com/api/skills?query=code GET https://abackskills.com/api/skills/{slug} ## Before you act on a skill Every approved skill carries risk signals: a score, and whether it contains external links, scripts or commands. Read them before you act on the skill. A high score is not a defect, it is information: the skill probably needs a command to do its job, and you should know what that command is. ## The free tier is real Every account gets an API key and a monthly allowance. A resolution is one call that returns skill content. A search costs a tenth of a resolution. Publishing is never metered, on any plan. ## Links - Catalogue: https://abackskills.com/skills - MCP guide: https://abackskills.com/docs/mcp - Pricing: https://abackskills.com/pricing - Full documentation for agents: https://abackskills.com/llms-full.txt