aislop CLI
A CLI that finds traces coding agents leave in code, like filler comments, swallowed errors and fake imports, and scores it 0–100
- 741 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The hook and agent mode run after agent edits, and the fix and agent modes change project files and may open a pull request.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/aislopDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The hook and agent mode run after agent edits, and the fix and agent modes change project files and may open a pull request.
This is third-party code. Review the repository files before installing.
What it does
aislop scans a project and looks for the template traces left by Claude Code, Cursor, Codex and OpenCode: comments above self-explanatory code, swallowed exceptions, hidden fallbacks, `as any` casts, hallucinated imports, duplicated helpers, dead code, todo stubs and oversized functions. It checks against more than 50 rules for 11 languages, including TypeScript, Python, Go, Rust, Java and C/C++. Every change gets a score from 0 to 100, and the scan is fast and deterministic with no LLM call. The quick start is `npx aislop@latest scan`. The `aislop fix` command repairs mechanical issues such as formatting, unused imports and dead code, and `aislop fix --claude`, `--codex` or `--cursor` hands the rest to an agent with full context. The `aislop agent` mode creates a local git worktree, runs a repair session with Codex, Claude Code or OpenCode, verifies the result with a rescan and optionally opens a draft pull request. `aislop hook install --claude` attaches a hook after every agent edit, hooks exist for Cursor, Gemini CLI and pi too, and rules are installed for Codex, Windsurf, Cline and Copilot. The `--quality-gate` mode blocks an edit if the score falls below the baseline. For CI there is `aislop ci` with JSON and SARIF, a GitHub Action and a pre-commit hook, and the `aislop-mcp` MCP server exposes the checks to Claude Desktop, Cursor and Codex.
Who it is for. Teams where agents write the code and reviewers who want a fast deterministic filter before a pull request instead of reading every diff by eye.
Good fit when
- An agent left comments that restate the code and swallowed exceptions in a diff and review does not catch them
- You need a CI barrier: a pull request fails if the score is below a set threshold
- An agent edit must not worsen the project score and the hook's quality-gate mode does that
- Mechanical clutter such as unused imports has to be cleaned with one command and the rest handed to an agent
- The project README needs a public badge with a code quality score
Not a fit when
- The project is in Swift, Kotlin or Scala: aislop supports 11 languages and withholds the score when they are missing
- You need deep vulnerability hunting: security rules are part of the set but do not replace a full scanner
- You need a semantic review of logic: the tool is deterministic and checks patterns rather than program meaning
- A public report is unacceptable: the `--share` command prints a link to a public report of the GitHub repository
Example request
Run aislop scan --changes, fix the mechanical findings with aislop fix and show what is left, with the score before and after.Limitations
The score reflects only the aislop rule set and does not prove the quality of the code as a whole. For projects whose main language is outside the supported list the tool does not score, so as not to distort the picture from a few incidental files.
Some checks use third-party programs such as Ruff and golangci-lint. They install with `aislop-tools`, and the core works without them. Installing the package does not run dependency scripts.
The `--share` command and the badge send the repository address to the scanaislop.com service, so do not use them for private projects. The `aislop agent` mode runs a local agent with write access to a git worktree, and the result reaches the project only after your `--apply` or `--pr`. The CLI license is MIT.
How to disable. Remove the hooks with `aislop hook uninstall --claude` and uninstall the package with `npm uninstall -g aislop` or `brew uninstall aislop`. Delete the `.aislop/config.yml` file in the project by hand.
FAQ
How do I scan a project without installing?
With `npx aislop@latest scan` in the project root. The scanner immediately gives a score from 0 to 100 and a list of findings.
How do I connect aislop to Claude Code?
Install the CLI and run `aislop hook install --claude`. The hook checks code after every agent edit and returns the findings.
Does aislop use a language model?
No, the scan is deterministic: the same code always gives the same score. A model joins only in the `aislop agent` and `aislop fix --claude` modes.
How do I set a CI barrier?
The `aislop ci` command applies the score threshold and exit code. The threshold is set with the `failBelow` key in `.aislop/config.yml`, and for PR checks add `--changes --base origin/<branch>`.
Which languages are supported?
TypeScript, JavaScript, Expo and React Native, Python, Go, Rust, Ruby, PHP, Java, C# and C/C++.
Discussion
No questions or reviews yet.
Related
CLI toolsA check for AI-generated code: empty functions, imports of things that do not exist, and docs that oversell.
A 0 to 10 score for CLAUDE.md and AGENTS.md: Reporails finds vague rules, repeats and broken imports
CLI installs a verify-by-running harness into Claude Code and Codex: build and tests, diff review and a separate QA agent.
Hooks for Claude Code and Codex CLI keep a ledger of agent actions and refuse a "done" claim until a check has passed.
An ESLint-style linter whose rules are judged by a model: it finds in agent-written code what syntax checks miss and sets a verdict for CI
Agent Reliability Analyzer statically scans AI agent code, finds reliability gaps with 340 plus rules and applies fixes
README in short
The README presents aislop as a trap for agents' template traces with a deterministic score. It gives a quick start, installation through npm, Homebrew and PyPI, reference for the scan, fix, agent and hook commands, rule and config setup, CI examples and links to the docs.