split-work
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes a custom directive
DISRESPEC-SPINEdesigned to override the agent's default communication style and output format. While used for optimization in this context, such behavioral overrides are characteristic of prompt injection techniques. - [PROMPT_INJECTION]: The skill processes untrusted, user-provided specifications and change plans to generate task packets. This creates an attack surface for indirect prompt injection where instructions embedded in the source data could influence the agent's behavior.
- Ingestion points: Input documents (specifications or change plans) in
SKILL.md. - Boundary markers: The skill does not define explicit delimiters to separate user input from instructional logic.
- Capability inventory: File system writes to
~/.agents/artifacts/across all scripts. - Sanitization: No validation or escaping of the requirement text extracted from source documents is performed.
Audit Metadata