create-tasks
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is documentation generation based on project context, which is performed without modifying existing source code.
- [SAFE]: Data access is scoped to the current repository for the purpose of identifying architectural patterns and implementation anchors, with no evidence of external exfiltration.
- [SAFE]: Write operations are strictly limited to the
docs/tasks/subdirectory, preventing interference with the codebase or system configuration. - [PROMPT_INJECTION]: The skill ingests data from PRDs and external documents, creating a potential surface for indirect prompt injection. This is mitigated by the skill's restricted output format (Markdown tasks) and a mandatory human clarification step.
- Ingestion points: PRD files, user prompt text, attachments, and linked documents.
- Boundary markers: The skill uses a 'Clarify before drafting' protocol to validate requirements and resolve ambiguities.
- Capability inventory: Read-only repository access and write access limited to the
docs/tasks/directory. - Sanitization: The workflow relies on user-mediated clarification and standardized output templates to manage content integrity.
Audit Metadata