openspec-new-change
Pass
Audited by Gen Agent Trust Hub on Mar 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local commands using the
openspecCLI tool to manage change workflows. This behavior is consistent with the skill's stated purpose of providing a structured interface for the CLI.- [PROMPT_INJECTION]: The skill includes an indirect prompt injection surface by processing user-provided descriptions to generate command arguments. Evidence chain: 1. Ingestion point: User input for change description (SKILL.md, Step 1). 2. Boundary markers: Absent. 3. Capability inventory: Execution ofopenspecCLI commands via shell. 4. Sanitization: Present; the skill instructs the agent to derive a kebab-case name from the input, which mitigates risk by restricting the allowed character set in the resulting command.
Audit Metadata