billing-core = { git = "...", branch = "main" } means the next cargo update builds whatever was last pushed to main, and one token = "..." line in .cargo/config.toml hands your registry token to everyone who can read the repository. This skill finds those and eight other Cargo and Rust mistakes before you build or release, and tells you how to fix each one.

Your agent runs a scanner on your project: Cargo.toml (every dependency table, target-specific tables, [workspace.dependencies], [patch] and [replace]), Cargo.lock, .cargo/config.toml and the older .cargo/config, credentials files, build.rs and every .rs file. It parses TOML with line numbers and reads Rust with a small lexer that knows nested comments, raw strings, char literals and lifetimes, so text in strings and comments never counts as code. It follows Cargo's own rules (workspace = true inheritance, Cargo.lock at the workspace root, config paths relative to the folder that holds .cargo) and reads .gitignore to tell whether Cargo.lock and credentials files are committed. It never runs Cargo, a build script or your code. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secrets are never printed.

What it checks

BLOCK (fix before shipping)

• C02: git dependencies, registry indexes, source replacements and Cargo.lock sources over http:// or git://

• C03: registry tokens in .cargo/config.toml or a credentials file in the project, passwords and tokens in URLs, and secrets in [env] or [package.metadata]

REVIEW (check before release)

• C01: git dependencies and patches with no rev or tag, so they follow a branch

• C11: "*" versions, >= requirements with no upper bound, and dependencies with no version at all

• C12: [patch], [replace] and paths overrides

• C13: binaries with no committed Cargo.lock (missing, or listed in .gitignore)

• C14: build.rs that starts other programs

• C15: build scripts that download: HTTP client crates in [build-dependencies], reqwest or ureq in build.rs, curl, wget or git clone

• C16: unsafe blocks, unsafe fn and unsafe impl, counted and listed per file

• C17: path dependencies, patches and vendored sources outside the repository

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• A risky and a safe example project, and the real report the scanner produces on the risky one

Limits

It reads your project's files as written. Your user-level ~/.cargo/config.toml, CARGO_* environment variables, the build scripts and code of your dependencies, procedural macros, and known vulnerabilities in crate versions (cargo audit covers those) are not checked. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 25 files: SKILL.md, the scanner in scripts/ (7 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (15 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.