The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local environment, creating a potential surface for indirect prompt injection. \n
Ingestion points: pome.json, .pome/link.json, agent system prompts, and source code (e.g., src/index.ts). \n
Boundary markers: Absent; there are no specific instructions for the agent to use delimiters or ignore embedded instructions within these files. \n
Capability inventory: Reading local files and using vendor-specific MCP tools (mcp__pome__*) for task listing, validation, and storage. \n
Sanitization: Absent; the skill does not specify any validation or sanitization steps for the ingested code or prompts. \n- [COMMAND_EXECUTION]: The documentation references the use of the pome CLI for agent registration and management. \n
Evidence: The skill instructions mention using pome register agent to confirm registration. \n- [EXTERNAL_DOWNLOADS]: The skill setup involves downloading components and connecting to services managed by the vendor. \n
Evidence: The skill is installed via npx skills add pome-sh/digital-twins and connects to the vendor's MCP server at https://mcp.pome.sh/mcp.

pome-suggest-tasks