work-item
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a significant surface for indirect prompt injection due to its core functionality of ingesting and acting upon untrusted data from the repository.
- Ingestion points: The skill reads and processes content from multiple sources including
plan.md, artifacts intasks/, user stories intasks/stories/, and canonical feature specs indocs/specs/(as seen inreferences/mode-refine.mdandreferences/mode-mockup.md). - Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the ingested content before it is processed by the agent or passed to sub-agents.
- Capability inventory: The skill has powerful capabilities including writing files to the disk, executing local Node.js scripts (e.g.,
export.cjs,open-report.cjs), and spawning multiple sub-agents (ui-ux-designer,general-purpose) using thespawn_agenttool. - Sanitization: There is no evidence of sanitization or filtering of the external content before it is interpolated into prompts or used to define sub-agent tasks.
- Risk: An attacker could place malicious instructions inside a task description or a feature spec which, when processed by this skill, could influence the behavior of the sub-agents or the execution of local scripts.
Audit Metadata