openspec-apply-change
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data that may contain instructions intended to influence or override agent behavior.
- Ingestion points: Data is ingested through the
contextandoperationGuidancefields of theopenspec instructionsJSON output, and from reading files specified in thecontextFilesarray. - Boundary markers: The skill does not define boundary markers or instruct the agent to ignore embedded instructions; instead, it explicitly directs the agent to follow the provided guidance.
- Capability inventory: The agent is authorized to execute
openspeccommands via Bash and to modify local files by updating task status checkboxes. - Sanitization: There is no evidence of data sanitization or validation of the content retrieved from the CLI or project files.
Audit Metadata