sdd-authoring
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a shell script
scripts/spec-quality-check.shused for validating the structure of markdown files. The script performs benign text searching usinggrepto ensure required headings (e.g., 'Problem Statement', 'User Stories') are present in the specification files. It does not execute arbitrary user input or perform risky system operations. - [REMOTE_CODE_EXECUTION]: No remote code execution patterns or external package installations were detected. All dependencies and references are local to the skill folder.
- [DATA_EXFILTRATION]: There are no network operations or access to sensitive file paths. The skill operates entirely on local specification documents.
- [PROMPT_INJECTION]: The instructions follow standard engineering documentation patterns. There are no attempts to bypass safety filters or override agent system prompts.
- [INDIRECT_PROMPT_INJECTION]: The skill processes requirement documents to generate specifications. While this is an ingestion surface, the skill includes 'Spec Quality Gates' and explicit 'Acceptance Criteria Design' guidelines that enforce structured validation, acting as natural boundaries for processed data. The capability tier is low (local file reading and basic shell validation).
Audit Metadata