openspec-create-spec-from-code
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the
openspeccommand-line tool to perform tasks such as version verification, listing existing specifications, and validating new specs. Command arguments like capability IDs are derived from the local environment and user input. - [INDIRECT_PROMPT_INJECTION]: The skill analyzes implementation and test files to determine intended system behavior. This introduces a surface where malicious instructions hidden in code comments or documentation could manipulate the agent's analysis or the resulting specification documents.
- Ingestion points: The skill reads source code, API schemas, and test files from the project directory through its subagent workflow.
- Boundary markers: There are no specific instructions or delimiters provided to ensure that instructions found within the code are ignored by the analyzing model.
- Capability inventory: The skill is capable of reading arbitrary project files, executing the
openspecCLI tool, and writing specification files to theopenspec/specs/directory. - Sanitization: The skill implements kebab-case normalization for capability identifiers, which acts as a basic safeguard for command arguments and file paths.
Audit Metadata