
Judges whether reproduction evidence proves a new regression or an exclusion
- Open source
- 8.2K GitHub stars
- Security

Judges whether reproduction evidence proves a new regression or an exclusion

Finds candidate regressions in a pinned code change by tracing its callers

Pins before and after source states and gathers PR or commit context for regressions

Coordinates multi-agent hunts for regressions in PRs and commits with staged verification

Independently re-runs a supplied proof-of-concept against before and after builds

Assembles an evidence-linked regression report from saved campaign decisions

Builds live before/after deployments and a portable PoC to reproduce a regression

Guides an agent through a security audit of Rust code focused on unsafe, FFI

Runs independent code reviews of changes through Codex or Antigravity CLIs

Guides the agent to write and test custom Semgrep rules for security and code patterns

Ports an existing Semgrep rule to other languages with tests for each variant

Reviews APIs and configs for misuse-prone designs and insecure defaults

Checks code against its specification to find requirements it breaks or omits

Guides an agent to build and run C/C++ or Rust code under AddressSanitizer to find memory

Guides an agent through multi-core AFL++ fuzzing of C/C++ projects, from build to triage

Guides an agent through coverage-guided fuzzing of Python code and C extensions

Sets up and runs cargo-fuzz harnesses to fuzz Rust crates and triage crashes

Measures which code a fuzzing campaign reaches and turns gaps into harness or seed work

Guides fuzzers with keyword and magic-byte dictionaries for libFuzzer and AFL++

Guides patching checksums and validation gates so a fuzzer can reach deeper code

Guides an agent in writing and fixing fuzzing harnesses for C, C++ and Rust targets

Guides building custom LibAFL fuzzers in Rust for targets standard fuzzers cannot handle

Guides an agent through building, running and triaging libFuzzer harnesses for C/C++ code

Guides an agent through enrolling a project in OSS-Fuzz and running its harnesses locally

Lets the agent set up and run a coverage-guided fuzzer against Ruby code and gems

Guides agents to test crypto implementations against Project Wycheproof known-attack

Overlays SARIF, weAudit and binary graph findings onto Trailmark code graphs

Triages mutation testing and necessist results into false positives, test gaps and fuzz

Builds call graphs and taint, blast-radius and entrypoint maps of code for security audits

Runs structural security checks on a branch or PR diff before merge