prompt-leverage

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests raw user prompts to generate refined instruction sets in SKILL.md and scripts/augment_prompt.py.\n
  • Ingestion points: The user prompt passed as an argument to scripts/augment_prompt.py and the prompts processed through the workflow in SKILL.md.\n
  • Boundary markers: The user-provided content is directly interpolated into the template's {normalized} placeholder without specific delimiters or instructions to prevent user input from overriding the framework's structure.\n
  • Capability inventory: The skill primarily generates text, but the resulting prompts are intended to be executed by agents that may possess file system and tool access.\n
  • Sanitization: The implementation applies basic whitespace normalization but lacks mechanisms to filter or escape potential injection keywords that could influence the final agent behavior.\n- [DYNAMIC_EXECUTION]: The test suite in scripts/test_augment_prompt.py uses dynamic loading to verify the functionality of the primary script.\n
  • Evidence: The script employs importlib.util and spec.loader.exec_module(module) to load and execute the local scripts/augment_prompt.py script at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 08:15 AM
Security Audit — agent-trust-hub — prompt-leverage