openspec-archive-change
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and follow behavior-influencing data (labeled as 'context' and 'operationGuidance') returned by the
openspecCLI tool. This pattern allows external content to define behavioral contracts for the agent during the execution of the workflow. - Ingestion points: In Step 1, the skill executes
openspec instructions archive --change "<name>" --jsonto fetch prompt-level behavior instructions. - Boundary markers: The instructions include logic to separate this external data from built-in steps, explicitly stating that resolved paths and CLI contracts must take precedence over the advisory guidance.
- Capability inventory: The skill possesses capabilities for file system modification (
mkdir,mv) and execution of theopenspecCLI tool suite. - Sanitization: The instructions require the agent to report conflicts rather than silently obeying external guidance and forbid the verbatim copying of this external data into project artifacts or archive summaries.
Audit Metadata