ai-launch-check

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill uses highly prescriptive instructions (e.g., 'Iron Law', 'Hard stop') to enforce specific behavior and decision-making criteria on the agent. While intended for procedural integrity, this rigid structure defines the core interaction logic.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its reliance on processing external files as evidence.
  • Ingestion points: The agent reads the contents of 'docs/ai-decisions/.md' and various evidence files, such as logs, code samples, and evaluation sets, provided by the user.
  • Boundary markers: There are no explicit instructions or delimiters defined to isolate the content of these files from the agent's internal instruction set, creating a risk that malicious instructions within these files could influence the agent's verdict.
  • Capability inventory: The skill is restricted to 'Read' and 'Write' tools. It does not possess capabilities for network communication or arbitrary command execution, which limits the potential impact of an injection.
  • Sanitization: No sanitization, filtering, or validation is performed on the content retrieved from the files before the agent processes it to verify checklist items.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 01:53 PM
Security Audit — agent-trust-hub — ai-launch-check