bridge-contract

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local script node scripts/spec-superflow.mjs state init <change-dir> to initialize state tracking for the generated contract. This is a standard project initialization task.
  • [DATA_EXFILTRATION]: The skill reads local markdown files (e.g., proposal.md, specs/, design.md) to extract requirements and constraints. This data access is restricted to the local environment and is necessary for the skill's purpose.
  • [PROMPT_INJECTION]: An indirect prompt injection surface is present because the skill parses potentially untrusted documentation content. Evidence: 1. Ingestion points: proposal.md, specs/, design.md, tasks.md. 2. Boundary markers: Absent. 3. Capability inventory: Execution of local maintenance scripts. 4. Sanitization: The skill implements a critical guardrail by requiring explicit user approval (DP-3) for the final contract before any further workflow steps occur.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 01:03 AM
Security Audit — agent-trust-hub — bridge-contract