Featured

Deploy OpenClaw in 60 seconds — 20% off logoDeploy OpenClaw in 60 seconds — 20% off

Launch OpenClaw on Hostinger in about 60 seconds and keep your agent live 24/7. Our referral link gives you 20% off, no coupon code needed.

Launch on Hostinger
Run your Hermes agent on Hostinger, fully managed logoRun your Hermes agent on Hostinger, fully managed

Launch Hermes on Hostinger in one click, fully managed, no VPS knowledge needed. Use code ZACAARON10 for 10% off.

Launch on Hostinger
Crawl and scrape any site into clean data, 10% off logoCrawl and scrape any site into clean data, 10% off

Firecrawl crawls and scrapes any site into clean markdown for your agent. Get 1,000 free credits, and new users get 10% off their first purchase.

Try Firecrawl free
Your own AI agent, running 24/7 with QwikClaw logoYour own AI agent, running 24/7 with QwikClaw

QwikClaw sets up and runs an always-on OpenClaw agent for you. One click, no config files, no server setup.

Deploy now
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you
SEO data APIs for your agent, $1 free credit logoSEO data APIs for your agent, $1 free credit

DataForSEO gives your agent live access to SERP results, keyword data, backlinks, and on-page SEO data through one API. New accounts get a $1 credit, good for up to 20,000 keyword or backlink lookups.

Try DataForSEO free
Reach 47,000+ AI builders

A flat monthly placement in front of developers actively installing AI tools. No lock-in, cancel anytime.

Advertise here
rust-ext-review-toolkit logo

rust-ext-review-toolkit

rust-ext-review-toolkit

code-qualityClaude Codeby clin1234

Summary

13 specialized agents and 4 commands for analyzing PyO3 Rust extensions. Tree-sitter-rust scripts detect unsafe-block soundness violations, panics crossing the FFI boundary, swallowed PyResults, #[pyclass] Send/Sync gaps, missing __traverse__/__clear__, and PyO3 version-migration debt. Includes a 'migrate' command for the Bound-API and free-threading migration checklist.

Install to Claude Code

/plugin install rust-ext-review-toolkit@rust-ext-review-toolkit

Run in Claude Code. Add the marketplace first with /plugin marketplace add clin1234/rust-ext-review-toolkit if you haven't already.

README.md

rust-ext-review-toolkit

A Claude Code plugin for reviewing Python extensions written in Rust with PyO3 — finding soundness bugs in unsafe code, PyO3 API misuse, GIL/lifetime discipline issues,

#[pyclass] trait-bound violations, panic-safety problems, and PyO3 version-migration debt.

Find the soundness bugs in your Rust extension before your users do.

What PyO3 prevents — and what it doesn't

PyO3's type system already eliminates most classic C-extension bug classes, so this toolkit does not scan for them:

  • Manual Py_INCREF/Py_DECREFPy<T> and Bound<'py, T> implement Drop
  • Most NULL-pointer dereferences — fallible APIs return PyResult<T> / Option<T>
  • Most use-after-free of Python objects — the 'py lifetime ties handles to the

interpreter

  • Memory safety in safe Rust — the borrow checker

What PyO3 does not prevent — and what this toolkit is for:

  • Soundness bugs inside unsafe (raw pyo3-ffi, transmute, manual buffer

protocol, C-library FFI)

  • #[pyclass] trait-bound violations — non-Send/Sync fields without

unsendable

  • Panics crossing the extern "C" FFI boundary
  • PyResult propagation gaps — .unwrap(), .ok(), let _ =, missing ?,

exception clobbering

  • Wrong handle kind — Bound vs Py vs Borrowed; legacy &PyAny
  • #[pymethods] protocol mistakes — missing __traverse__/__clear__,

__richcmp__ semantics

  • PyO3 version-migration debt and abi3 consistency
  • Free-threaded Python (3.13t/3.14t) readiness

Installation

Marketplace install

claude plugin marketplace add clin1234/rust-ext-review-toolkit
claude plugin install rust-ext-review-toolkit@rust-ext-review-toolkit

Try it without installing

git clone https://github.com/clin1234/rust-ext-review-toolkit.git
claude --plugin-dir rust-ext-review-toolkit/plugins/rust-ext-review-toolkit

Prerequisites

  • Claude Code installed and running
  • Python 3.10+ for the analysis scripts
  • tree-sitter and tree-sitter-rust: pip install tree-sitter tree-sitter-rust
  • Optional: a Rust toolchain with cargo clippy / cargo miri / cargo expand

for deeper cross-referencing

Quick start

In a PyO3 crate:

/rust-ext-review-toolkit:health      # Quick scored dashboard
/rust-ext-review-toolkit:hotspots    # Worst functions: unsafe + panic + complexity
/rust-ext-review-toolkit:explore     # Full analysis (all agents, phased)
/rust-ext-review-toolkit:migrate     # Bound-API / free-threading / abi3 checklist

Agents

13 agents — 11 script-backed, 1 qualitative, 1 preflight.

Preflight

  • rust-codegen-mapper — orientation guide: PyO3 version + features, file

classification, project-specific idioms; runs first so downstream agents triage accurately.

Safety-critical

  • unsafe-block-auditor — soundness of every unsafe block; raw pyo3-ffi

calls; transmute; manual buffer protocol (the #1 agent)

  • pyresult-propagation-checker.unwrap()/.ok() on PyResult, missing

?, exception clobbering

  • gil-discipline-checker — handles held across detach; nested attach;

foreign-callback GIL acquisition; free-threading declarations

  • panic-safety-checker — panic-prone calls reachable from #[pymethods] /

#[pyfunction]

PyO3-specific

  • pyclass-traits-checker#[pyclass] Send/Sync/frozen/unsendable
  • pyclass-protocol-checker#[pymethods] slot signatures,

__traverse__/__clear__, __richcmp__

  • module-init-checker#[pymodule] shape, module-level static state
  • lifetime-handle-checkerBound vs Py vs Borrowed; legacy &PyAny

Compatibility, quality, parity

  • pyo3-version-compat-scanner — deprecated/removed APIs by version; abi3

consistency

  • rust-complexity-analyzer — complexity scoring (includes unsafe-block

density)

  • git-history-analyzer — similar-bug detection, churn-based risk

prioritization

  • parity-checker — behavioral divergence between a Rust extension and its

pure-Python fallback

Commands

| Command | Purpose | |---------|---------| | explore | Full analysis, phased agent groups, selectable aspects | | health | Quick scored dashboard, all agents in summary mode | | hotspots | Worst functions first — unsafe + pyresult + panic + complexity | | migrate | Bound-API migration, free-threading readiness, abi3 — checklist output |

How it works

Tree-sitter-rust parsing. Scripts parse .rs source with tree-sitter-rust — style-agnostic and robust against macro-heavy code. Tree-sitter does not expand procedural macros, so #[pyclass]/#[pymethods] analysis is syntactic inference over the attribute + struct + method shape.

Crate discovery. Auto-detects PyO3 crates: Cargo.toml with a pyo3 dependency and a cdylib, pyproject.toml with maturin or setuptools-rust, or

setup.py with setuptools_rust.RustExtension. Resolves the PyO3 version, active features, Python floor, and free-threading intent.

Classification. Every finding is tagged:

| Tag | Meaning | |-----|---------| | FIX | Soundness bug, panic across FFI on a reachable path, error swallowing that surfaces wrong data | | CONSIDER | Likely improvement; PyO3 idiom migration; free-threading readiness gap | | POLICY | Maintainer's design call (adopt abi3? target free-threading? support subinterpreters?) | | ACCEPTABLE | Noted, no action — e.g. an unsafe block with a verified // SAFETY: comment |

External tools (optional). When a Rust toolchain is present, clippy /

miri / cargo-expand / cargo metadata are used opportunistically. Never required — Tree-sitter analysis is the baseline.

Limitations

  • Tree-sitter, not a compiler. Cannot expand procedural macros, resolve

trait-bound satisfaction, or do type inference. Scripts emit candidates (~20–40% false positives); agents confirm or dismiss each.

  • Macro-generated code is invisible. What #[pyclass] emits is inferred

from the source-level shape, not seen directly.

  • PyO3 evolves fast. API knowledge lives in version-keyed data files; the

analysis logic stays stable across PyO3 releases.

Sibling projects

| Project | Target | Parsing | |---------|--------|---------| | code-review-toolkit | Python source | ast | | cpython-review-toolkit | CPython runtime C | Regex | | cext-review-toolkit | C extensions | Tree-sitter-c | | rust-ext-review-toolkit | Rust/PyO3 extensions | Tree-sitter-rust |

Authors

Charlie Lin and Danzin.

License

MIT — see LICENSE.

Related plugins

Browse all →