pome-suggest-tasks
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes untrusted local source code (e.g., src/index.ts) and agent system prompts to generate task suggestions. Adversarial instructions within these files could potentially influence the agent's behavior during the analysis phase. 1. Ingestion points: Manifest files (pome.json, pome.yaml, pome.yml), source code (src/index.ts), and agent instructions (SKILL.md). 2. Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing external content. 3. Capability inventory: The skill interacts with the Pome control MCP server via mcp__pome__* tools (SKILL.md). 4. Sanitization: No explicit sanitization or filtering of the analyzed content is performed.
- [COMMAND_EXECUTION]: The workflow involves the use of the pome CLI tool for registration and management tasks like pome login and pome register agent (references/cold-walk.md).
- [DATA_EXFILTRATION]: Data derived from the analysis of local code and manifests is sent to the vendor's MCP server at mcp.pome.sh to synchronize with the team catalog (SKILL.md).
Audit Metadata