The Agent Skills Directory

[NO_CODE]: The skill consists entirely of Markdown documentation and JSON configuration examples. No executable scripts, binaries, or active code components are provided in the package.
[COMMAND_EXECUTION]: The documentation describes the standard functionality of MCP stdio servers, which involve spawning local child processes for tool execution. This is presented as an architectural pattern for local integration.
[REMOTE_CODE_EXECUTION]: Instructional examples include the use of npx to fetch and run MCP servers (e.g., @modelcontextprotocol/server-filesystem) from the official NPM registry. These are standard development workflows for the protocol.
[PROMPT_INJECTION]: The skill acknowledges the surface for indirect prompt injection when autonomous agents interact with external tools.
Ingestion points: Data retrieved from external services (e.g., Asana tasks, database queries) as described in references/tool-usage.md.
Boundary markers: The documentation recommends using explicit allowed-tools lists in command frontmatter to restrict the agent's capability surface.
Capability inventory: MCP servers can execute shell commands (stdio) or perform network operations (SSE, HTTP, WS).
Sanitization: The documentation focuses on configuration and authentication rather than input sanitization logic.

MCP Integration