The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and execute an external MCP server using the npx command.
Evidence: "command": "npx", "args": ["@thekeyword/mcp"] in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection by processing external data from API endpoints which could potentially contain malicious instructions.
Ingestion points: GET /api/v1/settings and GET /api/v1/usage in SKILL.md.
Boundary markers: No explicit boundary markers or delimiters for API content are provided in the instructions.
Capability inventory: The skill uses an MCP server and HTTP requests to interact with the external service.
Sanitization: No sanitization or validation of the retrieved API data is mentioned in the skill instructions.

thekeyword-setup