Rustpy Review Toolkit — AI skill for Claude Code
Claude Code plugin that statically reviews the RustPython interpreter's own Rust source for Python-reachable panics, object-model unsafe-soundness bugs, and GC traverse-completeness.
How to install Rustpy Review Toolkit
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open ReviewToolkits/rustpy-review-toolkit and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Rustpy Review Toolkit does
Claude Code plugin that statically reviews the RustPython interpreter's own Rust source for Python-reachable panics, object-model unsafe-soundness bugs, and GC traverse-completeness. Tree-sitter-rust-powered, static-first.
Alternatives in AI
- Jcodemunch MCP — Cut AI token costs 95%+ on code exploration 2.6k ★
- Azure Vision Java — Image captioning, OCR, and object detection 1.8k ★
- Loushang — AI-native agent harness for coding workflows by python: multi-model LLM orchestration, stateful sessions, tool 1.3k ★
README
rustpy-review-toolkit
A [Claude Code](https://docs.anthropic.com/en/docs/claude-code) plugin that statically reviews the **[RustPython](https://github.com/RustPython/RustPython) interpreter's own Rust source** for soundness and crash bugs — from the runtime-implementer's perspective.
RustPython is a Python interpreter written in Rust. This toolkit reviews *the interpreter itself*, not extensions built with it, and **not PyO3**. RustPython shares PyO3's vocabulary (`PyRef`, `PyResult`, `#[pyclass]`) but has its own object model, cycle-collecting GC, bit-packed `RefCount`, and `threading`-feature toggle — so every check is RustPython-native, not transplanted from a PyO3 toolkit.
What it finds
- Python-reachable panics (the flagship) —
.unwrap()/.expect()/panic!/ index / arity sites in the VM, tiered by reachability (py>protocol>internal) so a crash a Python program can trigger surfaces while internal invariants stay quiet. This is RustPython's dominant confirmed crash class. - Unsafe soundness (the crown jewel) — cross-method pointer-cast inconsistency in the object model (the
PyAtomicRefSIGSEGV shape) and handle-type transmute without arepr(transparent)/TransmuteFromObjectguard. - GC Traverse-completeness (first-class, experimental) —
#[pyclass]payloads that own Python references but declare notraverse,#[pytraverse(skip)]on ref-owning fields, and manualtraversebodies that miss a field. - Complexity and git history — reused, RustPython-calibrated.
Scanners find *candidates* (deliberately high recall); the agents confirm or dismiss each by reading the real code and judging Python-reachability.
Install
claude plugin marketplace add devdanzin/rustpy-review-toolkit
claude plugin install rustpy-review-toolkit@rustpy-review-toolkit
Commands
| Command | What it does |
|---|---|
/rustpy-review-toolkit:explore [scope] [aspects] |
Full phased review: discovery → preflight orientati |
Related Skills
Semble Rs
Fast, AI-agent-native code search in Rust — hybrid BM25 + semantic, Tree-sitter AST chunking, dependency & imp
Fast File Search
Ultra-fast file search toolkit for AI agents, Neovim, and code editors. MCP server for Claude Code/Cursor/Code
Autonomous Code Review Agent
A production-style AI code review platform that uses Claude to analyze uploaded source code, identify potentia
Ast Outline
ast-outline lets AI coding agents pull exactly the code context they need — a repo skeleton, a file outline, o
Reporecall
Local codebase memory for Claude Code. Tree-sitter AST indexing (22 languages), hybrid keyword + vector search
Rpg Explainer
CLI tool to analyze IBM RPG programs using Tree-sitter and Claude LLM
Related Agents
Interface Verifier
Validates interfaces compile correctly across multiple languages. Checks for type safety, circular dependencie
Brief Reviewer
Statically reviews a design brief or requirement (DECISION-BRIEF.md, a spec's Problem/Context section, or an e
P Code Review Security
Invoked by p-code-review to run the security channel across the entire working tree (no diff). Adversarial rea