skills/noih/dev-skills/sdd/Gen Agent Trust Hub

sdd

Pass

Audited by Gen Agent Trust Hub on Nov 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from spec artifacts (e.g., proposal.md, plan.md) and project files. * Ingestion points: Reads requirements and plan files which could contain malicious instructions. * Boundary markers: Employs delimiters to wrap verbatim spec text when passing data to other tools. * Capability inventory: Capable of writing to project files and executing shell-based test runners. * Sanitization: Includes strict instructions to treat spec text as evidence only and to ignore any embedded control flow or permission overrides.
  • [COMMAND_EXECUTION]: The skill invokes project-specific test runners and build tools (e.g., npm, cargo, pytest). It mitigates the risk of running malicious local scripts by requiring the agent to inspect command definitions and configuration in the repository before execution.
  • [EXTERNAL_DOWNLOADS]: References external companion skills such as grill-me and noih/dev-skills. The skill explicitly states it will never auto-install, fetch, or execute remote code, requiring manual user intervention for any external dependencies.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 18, 2026, 05:35 PM