SkillsPicker

What the SkillsPicker MCP server does

The SkillsPicker MCP server helps agents find relevant open-license skills for a project. Its recommendations are based on two kinds of tags: the technologies in the project stack and the task the agent needs to perform. The results refer to agent skills represented by SKILL.md files.

This makes the service relevant when an agent needs to identify reusable instructions or capabilities for a particular development context. For example, a project’s technology tags and a task’s tags can be used as the basis for finding matching skills. The available material does not describe ranking rules, result fields, or the source catalog beyond the open-license and SKILL.md characteristics.

How it works

SkillsPicker is remote and read-only. An MCP client can use it to request recommendations without giving the service permission to change project files or other resources. The service does not require authentication, so the provided description does not indicate any API key, account, or credential setup.

The matching inputs are a project’s stack tags and task tags. The service then recommends skills associated with those inputs. The available information does not specify whether tags are inferred automatically, supplied by the client, or obtained through another workflow, so implementations should not assume a particular project-discovery method.

Tools and capabilities

The documented capability is skill recommendation. In practical terms, the SkillsPicker MCP server is intended to help an agent discover open-license SKILL.md skills that fit a project and its current task.

Supported characteristics include:

• Recommendations based on stack tags.
• Recommendations based on task tags.
• Open-license agent skills as the recommendation target.
• Read-only remote access.
• No authentication requirement.

No individual MCP tool names, parameters, response schema, pagination behavior, or filtering options are provided. Clients should therefore treat the published description as the capability boundary rather than expecting undocumented operations.

Limitations and notes

The service is described as read-only, so it should be evaluated as a discovery or recommendation layer rather than a system for installing, editing, or managing skills. The available material does not state that it installs returned skills, modifies a repository, validates licenses, executes SKILL.md files, or analyzes source code.

The description also does not identify supported MCP clients, transport details, an installation command, configuration settings, rate limits, or an uptime policy. Those details should be confirmed from the service documentation before deployment. Because the SkillsPicker MCP server is remote and does not require authentication, teams should still review how recommendation requests are sent and what project or task tags they choose to provide.