spec-as-source
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides workflow guidelines and templates for BDD. No malicious code, obfuscation, or unauthorized access patterns were identified.\n- [COMMAND_EXECUTION]: The workflow involves running
git status --porcelainto maintain isolation between specification and code zones, which is a standard and safe local operation for development-oriented tasks.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes instructions to read project configuration and specifications to guide the workflow.\n - Ingestion points: Reads
openspec/config.yamlandspec.md(viaSKILL.mdandreferences/tasks.md) to identify the technology stack and draft implementation tasks.\n - Boundary markers: None present; the skill treats the project-local configuration as trusted data.\n
- Capability inventory: Capability is limited to drafting local Markdown documentation and task files. No network operations or dangerous subprocess calls are authorized or present.\n
- Sanitization: None; the agent relies on the local, user-controlled nature of the project repository.
Audit Metadata