any-percent

Pass

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface where external, untrusted content from design documents is used to drive agent behavior.
  • Ingestion points: Ingests context from docs/plans//design.md and user-supplied answers to constraints questions.
  • Boundary markers: Missing explicit delimiters or warnings to ignore instructions embedded within the design context files.
  • Capability inventory: The skill uses tools to write files (mcp__speed-run__generate_and_write_files), edit existing code, and execute shell commands via git and test runners.
  • Sanitization: No sanitization or validation of the ingested design markdown is performed before interpolation into agent prompts.
  • [COMMAND_EXECUTION]: The skill executes shell commands for repository management and automated testing of generated code.
  • Evidence: Uses git worktree commands to create and remove implementation environments.
  • Evidence: Instructs the agent to execute code during the 'Run tests' phase of implementation variants.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 16, 2026, 05:23 AM
Security Audit — agent-trust-hub — any-percent