kanban-worker
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted task data and project-level documentation to guide agent actions without protective delimiters.
- Ingestion points: According to
SKILL.md, the agent readstitle,description,tags, andreviewFeedbackfrom the Light-Kanban API (GET /api/tasks), and project-specific instructions from files likeAGENTS.md,CLAUDE.md, andREADMEin the workspace. - Boundary markers: The instructions lack requirements for the agent to wrap external task content in delimiters or use "ignore embedded instructions" warnings during processing.
- Capability inventory: The skill empowers the agent to "execute the task inside the project's own workflow," which implies high-privilege access to shell commands and file system operations.
- Sanitization: There is no evidence of input validation or sanitization for the retrieved task content or project documents.
Audit Metadata