Fabraix Red Team Playground
Free · Open source
AI security researchers
Argot
FreeOpen SourceSelf-Hosted
Your AI coding agent passes all the lints, all the type checks, and all the tests — then silently ships a Django import into a FastAPI repo, or disables a rate-limit test because the middleware it was supposed to fix is too hard to touch. Argot exists to catch that class of problem before it merges.
Similar tools
Fabraix Red Team PlaygroundFreeFabraix Red Team Playground is an open-source, MIT-licensed testing environment structured…
EclusiaFreemiumEclusia runs locally on your machine and replaces names, IBANs, phone numbers, API keys, and…
Patronus ScannerFreemiumPatronus AI positions itself as evaluation and monitoring infrastructure for production LLM…
NeuralTrust TrustGatePaidNeuralTrust TrustGate sits between your agents and the models, tools, and services they reach…
Get This Tool
License: License: unverified
Local-run terms: Install via npm, run argot init locally to build snapshot, commit snapshot for shared use; checks run against committed baseline.
Summary
Argot runs statistical checks against your repository's own commit history to flag code that looks foreign to the codebase — wrong libraries, reversed import direction, misplaced logic, or tests quietly skipped to cover up a failing change. It checks a diff in roughly 200ms with no network call and no second LLM in the loop. The vendor reports 97.3% of foreign patterns caught against a 0.25% false-positive rate on real edits. The ceiling is detection scope: it surfaces what statistically doesn't fit, but it cannot enforce rules that have never appeared in your history. Repos with sparse or inconsistent histories get weaker signals.
Bottom line: Argot fits any CI pipeline auditing AI-generated PRs in a repo with consistent conventions; it does not replace code review for repos whose history is too thin or too inconsistent to train a reliable baseline.
Find your fit
Before you choose Argot
More options in Guardrails & Safety. Compare pricing, platforms, and features side by side.
Eclusia
Freemium
Legal professionals handling confidential cases
Pros
- Checks a diff in roughly 200ms with no external model call, so CI feedback is available inside the same pipeline stage as your linter — not after a separate async job that blocks the queue.
- Flags foreign library usage by cross-referencing your repo's actual import history, which means a Django import in a FastAPI codebase fails the check even when it compiles and passes type-checking cleanly.
- Pairs disabled tests with the production code they cover in the same flag, so the 'all tests pass' green check no longer hides a skip marker an agent added to avoid fixing a rate limiter.
- Runs entirely local with no API dependency, which means no credentials to rotate, no usage caps during a deployment spike, and no code leaving the machine — a meaningful constraint for teams under data-handling policies.
- Cites the specific commits behind each flag, so a reviewer can reproduce the finding in the repo's own history rather than trusting a score with no traceable evidence.
Cons
- Repos with sparse commit history or mid-migration codebases — actively moving from requests to httpx, say — will see degraded signal because the baseline reflects a transitional state; teams in that window must manually tune thresholds or suppress whole detector categories until the migration stabilizes.
- There is no API surface, so connecting Argot to anything beyond a CLI invocation requires the team to own the integration glue; teams building a multi-tool review pipeline with webhook triggers or programmatic result ingestion are writing that plumbing themselves.
- Argot catches what statistically diverges from history — it cannot enforce a rule that has never appeared in the codebase; teams whose architecture conventions exist only in a wiki, not in merged commits, will not see those conventions reflected in flags, and will eventually move to a rule-based linter or a tool that accepts explicit policy definitions as input.
About
- Platforms
- CLI, npm, GitHub Actions
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- August 14, 2026
Best For
Who it's for
- Teams using AI coding agents
- Repositories with strict style and architecture rules
- CI pipelines needing fast, deterministic checks
What it does well
- Auditing AI-written PRs for repository fit
- Preventing test weakening in agent-generated code
- Enforcing internal layering and import conventions
Integrations
Frequently Asked Questions
- Is Argot free?
- Argot offers a free option. Hosting, model usage, or optional services may have separate costs; check the pricing details and official terms.
- Is Argot open source?
- Yes. Argot is open source.
- Can I self-host Argot?
- Yes. Argot supports self-hosting on your own infrastructure.
- What platforms does Argot support?
- Argot is available on: CLI, npm, GitHub Actions.
Curated lists that include this category
When an AI agent slips a foreign import past every check
Your AI coding agent passes all the lints, all the type checks, and all the tests — then silently ships a Django import into a FastAPI repo, or disables a rate-limit test because the middleware it was supposed to fix is too hard to touch. Argot exists to catch that class of problem before it merges.
How Argot checks code
Argot runs statistical checks against your repository’s own commit history to flag code that looks foreign to the codebase — wrong libraries, reversed import direction, misplaced logic, or tests quietly skipped to cover up a failing change. It checks a diff in roughly 200ms with no network call and no second LLM in the loop. The vendor reports 97.3% of foreign patterns caught against a 0.25% false-positive rate on real edits. The ceiling is detection scope: it surfaces what statistically doesn’t fit, but it cannot enforce rules that have never appeared in your history.
Platforms and limits
Argot runs on CLI, npm, and GitHub Actions with integrations for Claude plugin, Git hooks, and CI. A self-hosted option is available. Repos with sparse commit history or mid-migration codebases see degraded signal because the baseline reflects a transitional state.
Who it is for / who should skip it
Teams using AI coding agents, repositories with strict style and architecture rules, and CI pipelines needing fast, deterministic checks gain the most. Teams in active migrations must manually tune thresholds or suppress detectors until the history stabilizes.
Contribute
- Work on Argot? Claim this listing.
- Embed a badge for your README or site.
Sign in to rate this tool or suggest an edit.