Awesome Reviewers

Verification · Code review

LLM application security review

Reviews code that calls models for the OWASP Top 10 for LLM applications: prompt injection, output handling, excessive agency, unbounded consumption.

Sub-techniques

  • Static review of LLM call sites V1039.001

    Check how untrusted text reaches prompts and how outputs reach tools or HTML.

  • Red-team probing V1039.002

    Send adversarial prompts to a running application and score the responses.

  • Runtime input and output scanning V1039.003

    Scan prompts and responses in production for injection and leakage.

How to build it

  1. List every place untrusted text enters a prompt and every place model output is executed or rendered.
  2. Run a red-team suite against those paths in CI.
  3. Cap tokens, tool calls and spend per request.

Signals in a repository

  • promptfoo or garak config, prompt templates, and tool definitions in the repository.