skills/trailofbits/skills/atheris/Gen Agent Trust Hub

atheris

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The Dockerfile configuration includes downloading the LLVM GPG key and repository configuration from apt.llvm.org. This is a standard procedure for installing recent LLVM/Clang versions required for fuzzer instrumentation.
  • [INDIRECT_PROMPT_INJECTION]: The skill documents a surface for processing untrusted data via fuzzer harnesses. 1. Ingestion points: TestOneInput(data: bytes) in SKILL.md and examples.md. 2. Boundary markers: Absent, though FuzzedDataProvider is used for structured input. 3. Capability inventory: Execution of target functions like json.loads and HTTPResponse.read with fuzzed data. 4. Sanitization: Recommended use of try-except blocks to catch expected exceptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 01:16 PM
Security Audit — agent-trust-hub — atheris