openspec-new-change
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the openspec CLI tool (e.g.,
openspec new change,openspec status, andopenspec instructions). - [INDIRECT_PROMPT_INJECTION]: The skill ingests user descriptions and interpolates derived strings into shell commands, creating a potential injection surface.
- Ingestion points: User input collected via the
AskUserQuestiontool to describe a feature or fix in Step 1. - Boundary markers: The instructions rely on wrapping the interpolated
<name>variable in double quotes within the bash commands, but lack specific escape character handling. - Capability inventory: Shell execution of CLI tools for creating directories and fetching artifact templates as described in SKILL.md.
- Sanitization: The skill provides a specific instruction to derive a kebab-case name and validate it before execution, which serves as a basic input filter.
Audit Metadata