pome
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data in the form of managed-agent YAML definitions and local repository files to determine which workflow skill to route the user to.
- Ingestion points: Ingests Claude managed-agent YAML and local repository agent manifests to determine routing (SKILL.md).
- Boundary markers: None explicitly defined for the ingested context.
- Capability inventory: Controls workflow via MCP tools (run_task, register_agent, save_task) and the pome CLI (SKILL.md).
- Sanitization: No specific sanitization or validation of the input YAML or file content is described.
- [EXTERNAL_DOWNLOADS]: The skill references the vendor's official MCP server at mcp.pome.sh and a web dashboard at app.pome.sh. These are standard components of the vendor's ecosystem and are necessary for the skill's primary functionality.
Audit Metadata