The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the 'agentation' package using the project's package manager and uses 'npx' to execute utilities like 'agentation-mcp'. These resources are fetched from the NPM registry.
[COMMAND_EXECUTION]: The skill uses several shell commands to configure the development environment and the agent's Model Context Protocol (MCP) settings, including:
npx -y agentation-mcp server to launch the feedback server.
npx add-mcp or claude mcp add to persist the server configuration within the agent's system.
npx agentation-mcp doctor for diagnostic verification.
[INDIRECT_PROMPT_INJECTION]: The skill establishes a feedback loop where the agent is instructed to read annotations and comments provided by a user through a browser-based toolbar. This creates a surface where untrusted data from the browser is ingested into the agent's context.
Ingestion points: The agent retrieves user-generated annotations via the agentation_get_all_pending tool or similar session-scoped interfaces.
Boundary markers: No specific delimiters or safety instructions are provided to the agent to treat the retrieved annotation data as untrusted or to ignore potential instructions embedded within them.
Capability inventory: The skill possesses the capability to execute shell commands (npx), write to project files (Next.js layout files), and modify the agent's persistent MCP configuration.
Sanitization: The instructions do not specify any sanitization, filtering, or escaping for the content retrieved from the browser annotations before the agent processes it.

agentation