pipa-problem-framing

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted external data, such as client briefs, stakeholder notes, and tracker feedback, which could contain adversarial instructions.
  • Ingestion points: Workflow steps 2 and 3 in SKILL.md describe reading from briefs, kickoff notes, and live app feedback from platforms like Slack and Notion.
  • Boundary markers: The skill enforces a specific Markdown 'Output Contract' to structure and delimit findings.
  • Capability inventory: The skill uses composio-mcp for discovery and interaction with Notion, Linear, Slack, and Jira, and reads configuration from local files (~/.pipa/).
  • Sanitization: The instructions explicitly require the agent to distinguish between source facts and inferences, and strictly mandate explicit user approval before any external write operations occur.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 08:05 AM
Security Audit — agent-trust-hub — pipa-problem-framing