Skip to content

Releases: ars-engine/ars

Release list

Security release. If you use 0.1.0, upgrade.

Security release. If you use 0.1.0, upgrade. Latest
Latest

Choose a tag to compare

View all tags
@waspa waspa released this 28 Oct 17:35
v0.1.1
cedef0a
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

This release fixes the issues in GHSA-xhg7-866f-6fcp, which were reported privately and reviewed by @crenshawdev. Thank you for the careful report, and for reviewing the fix twice.

In 0.1.0, analyzing a repository you don't trust, or visiting a malicious website while ARS was running, could run commands or read files on your machine. 0.1.1 closes those paths.

What changed

  • Analysis treats a repository as data. Prettier configs are read as JSON, YAML or TOML and never executed. Git no longer runs inside an analyzed folder, so a repository's .git/config can't run commands.
  • The local API is locked down. Every request needs a session token that the UI fetches automatically, and requests from other websites are refused.
  • Only regular files inside the repository are read. Symlinks, special files and paths that lead outside the repository are skipped.
  • Lab agents are confined. File tools stay inside the run's workspace, and Bash runs sandboxed. If the sandbox can't start (on Linux it needs bubblewrap and socat), the run fails instead of running unsandboxed. Agents get a minimal environment and no MCP servers.
  • Pattern rules run with a time limit, so a bad regex can't freeze the server.
  • Git URLs with credentials in them are refused. ARS clones with your machine's own git credentials.

SECURITY.md describes every protection and the risks that remain.

Upgrading

git pull
npm install
npm run dev

Stop any ARS server that is still running from 0.1.0 first, or the new one can't use port 4317.

Also in this release

  • The docs say that ARS supports React projects only (TypeScript or JavaScript).
  • The docs explain the goal: install your skills into new projects, so Claude Code builds them your way from the first file.
Assets 2
Loading

v0.1.0 — first public release

v0.1.0 — first public release Pre-release
Pre-release

Choose a tag to compare

View all tags
@waspa waspa released this 26 Oct 17:38
v0.1.0
3670c1c

The first public version of ARS, an engine that learns how you write code from one of your projects, turns it into Claude Code skills, and measures on unseen tasks whether a coding agent follows them.

This is an early release. Expect rough edges, and please open an issue when a rule ARS learns from your code is wrong.

Highlights

  • Signature in six layers: formatting, syntax idioms, naming, component anatomy, architecture and voice. Every rule links to the files it was learned from, with counterexamples, and is self-tested against a planted violation.
  • Review inbox: approve, reject or reword each rule. Only approved rules reach the skills and the scores.
  • Versioned skill packages: SKILL.md files, de-domained exemplars, a formatter config, and an edit hook that formats each file and blocks off-style edits while the agent works. Install them from the UI or with ars install.
  • Lab: reconstruction (T1), transfer to new apps (T2) and indistinguishability (T3) tests, each run under three conditions (no skills, a length-matched generic guide, your skills), with repeats and bootstrap confidence intervals.
  • Improvement loop: diagnoses why a rule failed, proposes a fix, and promotes it only if it passes a regression gate on held-out tasks.
  • CLI: ars analyze, ars check, ars install and the Claude Code hook.

Requirements

  • Node 22.13 or newer, git and npm.
  • Analysis, review and skill generation need nothing else.
  • Deep-read and judges need an Anthropic API key. Agent runs need an API key or a Claude Code login.

Getting started

git clone https://github.com/ars-engine/ars.git && cd ars
npm install
npm run dev   # UI on http://localhost:5317

See Getting started and Costs before running Lab experiments, and Security for what leaves your machine.

Known limitations

  • Deep analysis covers TypeScript, JavaScript and React only. Other languages get a basic inventory.
  • There is no automated test suite yet.
  • The T3 "spot the real file" game can pick runs from any condition, not only the ARS one.
  • The Lab verdict compares means, not confidence intervals, so small runs can pass or fail by noise.
  • The baseline probe is only available after a skill version exists.

What's next

  • Codex and other coding agents, compared side by side in the same Lab.
  • Deep analysis for more languages.
  • Automated tests and CI.
Loading