ai-work-governance
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for AI agents to ingest untrusted data from external sources such as support tickets and customer logs to autonomously create GitHub issues and pull requests. This configuration creates an attack surface for indirect prompt injection.
- Ingestion points: Processes external customer conversation, support ticket, or production log data to generate work evidence in the Filing Logic and Handling Customer Evidence sections of SKILL.md.
- Boundary markers: The skill suggests using schema-based gates and agent self-checks but explicitly notes that automated quality validation of field content is unreliable and that human review gates often erode.
- Capability inventory: Autonomous creation and filing of GitHub issues, tasks, and pull requests without per-item human review.
- Sanitization: Recommends automated regex-based scrubbing for PII and credentials, along with content generalization to remove specific identifiers, though acknowledges these are best-effort controls.
Audit Metadata