sdd-draft-design

Pass

Audited by Gen Agent Trust Hub on Jul 12, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from the local codebase and external specifications.
  • Ingestion points: The skill reads 'the real codebase' and 'proposal/specs' files to inform design (SKILL.md).
  • Boundary markers: There are no defined delimiters or instructions provided to the agent to treat ingested data as non-executable content.
  • Capability inventory: The skill is limited to file system reads and the creation of Markdown design files; it explicitly forbids code edits, builds, or system state changes (SKILL.md).
  • Sanitization: No evidence of input validation or content escaping is present in the instructions.
  • [SAFE]: The skill implements strong administrative controls, including a 'Plan-only' hard rule and mandatory 'Decision Gates' that require explicit user approval before any output is committed to the file system.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 12, 2026, 05:26 PM
Security Audit — agent-trust-hub — sdd-draft-design