pipa-roadmap
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or persistence mechanisms were detected in the skill instructions or associated files.
- [DATA_EXPOSURE]: The skill accesses configuration files located in
~/.pipa/for style and connector preferences. These are application-specific files and do not include sensitive system credentials such as SSH keys or cloud provider secrets. - [PROMPT_INJECTION]: Instructions are focused on structured workflow execution and specifically include a rule to ignore presentation-layer instructions if they conflict with safety or approval gates.
- [DATA_EXFILTRATION]: While the skill interacts with external services, all network operations are performed via discovery tools and are subject to a mandatory human-in-the-loop approval process for any state-changing actions.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, creating a potential injection surface.
- Ingestion points: External data from Linear, Notion, Jira, and Drive (referenced in
evals.json). - Boundary markers: No explicit delimiters for ingested data are defined in the instructions.
- Capability inventory: The skill is capable of performing external roadmap and tracker writes via tools.
- Sanitization: There is no explicit sanitization of tool output described.
- Mitigation: The risk is mitigated by the safety protocol requiring the agent to display the exact change and obtain explicit user approval before any write operation.
Audit Metadata