The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The instructions direct the agent to install a remote MCP server from https://mcp.pome.sh/mcp using the claude mcp add command. This pattern involves executing code hosted on a remote server controlled by the vendor.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands locally, specifically the pome CLI for agent registration (pome register agent). It also references a REST launcher mechanism used to spawn external processes.
[EXTERNAL_DOWNLOADS]: The skill relies on external resources hosted on the vendor's domains, including mcp.pome.sh for the MCP control plane and app.pome.sh for reporting dashboards.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data, specifically the user's agent prompts and source code, to suggest testing tasks.
Ingestion points: Agent manifest files (pome.json), project repository code, and agent system prompts (SKILL.md).
Boundary markers: None identified; the skill does not specify delimiters to separate user data from instructions.
Capability inventory: The skill has access to shell command execution (pome CLI), process spawning (REST launcher), and network communication via the Pome MCP.
Sanitization: No explicit sanitization or filtering of the ingested agent code or prompt content is described.

pome