openspec-audit-change
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-authored documentation files, creating a potential surface for indirect prompt injection.\n
- Ingestion points: The skill reads
proposal.md,design.md,tasks.md, and specification files (specs/**/spec.md) from the repository to perform its audit.\n - Boundary markers: Audit logic is isolated within a subagent, and the skill requires explicit user confirmation before applying any automated 'autofixes' to files.\n
- Capability inventory: The skill has the capability to write to the filesystem (audit reports) and perform modifications on existing project artifacts.\n
- Sanitization: The instructions do not specify a content sanitization or validation process for the artifact data before it is processed by the auditor.\n- [COMMAND_EXECUTION]: The skill relies on the
openspecCLI to manage change status and metadata. This tool usage is central to the skill's purpose and facilitates the retrieval of structured JSON data for the audit process.
Audit Metadata