Fabraix Red Team Playground
Free · Open source
AI security researchers
ButterClaw
FreeOpen SourceAPISelf-Hosted
Authorized AI agents that behave correctly during review have a habit of drifting — pulling credentials they shouldn't touch, leaking context through indirect prompts — and most observability stacks have no idea it's happening. ButterClaw exists to catch that drift before it becomes an incident.
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…
ArgotFreeArgot runs statistical checks against your repository's own commit history to flag code that…
Get This Tool
License: Apache-2.0 Any use incl. commercial
Local-run terms: Users can clone the repo, run via Docker Compose, and use under Apache 2.0 terms for commercial purposes.
Screenshots 4
Pricing
- Model
- Free
Summary
ButterClaw runs as a self-hosted security monitor for AI agent deployments, watching post-authorization behavior for prompt injection attempts, credential exfiltration, and policy violations in real time. When it detects a breach condition, it can automatically revoke OAuth tokens and shred local credentials without waiting for a human to intervene. All of this runs locally — no cloud telemetry, no outbound data — which matters when the agents themselves are handling sensitive credentials. The Apache 2.0 license and Docker deployment model mean you own the stack entirely. The trade-off: without a vendor-hosted control plane, your team carries the operational weight of tuning and maintaining policies.
Bottom line: The right call for a self-hosted agent deployment where credential exposure is the primary risk — and a poor fit for teams without the engineering bandwidth to own security policy configuration and incident response infrastructure themselves.
Find your fit
Before you choose ButterClaw
More options in Guardrails & Safety. Compare pricing, platforms, and features side by side.
Eclusia
Freemium
Legal professionals handling confidential cases
Pros
- Automatic credential revocation on breach detection — OAuth tokens and local credentials are shredded without waiting for human review — so a compromised agent cannot persist access while an alert sits in a queue.
- Full local execution with no outbound telemetry, which means the security monitoring layer itself does not become an exfiltration vector for the sensitive credentials it is watching.
- Adjustable security posture from passive observation to active remediation, so teams can validate detection accuracy before enabling automated token revocation in production.
- Apache 2.0 license with Docker deployment, which means no vendor lock-in and no licensing cost as deployment scale grows.
- API availability, so breach events and policy state can be piped into existing incident response or SIEM tooling rather than treated as a standalone silo.
Cons
- Policy tuning is entirely the operator's responsibility — when the detection model flags legitimate agent behavior as a breach and begins revoking credentials, there is no vendor support path, only internal rollback and reconfiguration. Teams without a dedicated security engineering function hit this wall fast.
- The self-hosted model scales with your infrastructure investment, not a vendor's. High-throughput deployments with dozens of concurrent agents require the operator to provision, monitor, and maintain the ButterClaw stack itself — at the point where operational overhead exceeds the engineering team's capacity, teams move to a managed agent security vendor with SLA-backed availability.
- No described multi-agent coordination or cross-agent behavioral correlation in the scraped source material. Teams monitoring complex pipelines where one agent's output feeds another may find per-agent monitoring insufficient for detecting multi-hop exfiltration chains.
About
- Platforms
- Docker, Linux
- API Available
- Yes
- Self-Hosted
- Yes
- Last Updated
- August 14, 2026
Best For
Who it's for
- Self-hosted agentic AI deployments handling sensitive credentials
- Users needing adjustable security levels from observation to active token revocation
What it does well
- Monitoring authorized AI agents for behavioral drift after initial authorization
- Detecting indirect prompt injections and exfiltration attempts in real time
- Automatically revoking OAuth tokens and shredding local credentials on breach detection
- Running local security policies without cloud telemetry or outbound data
Integrations
Frequently Asked Questions
- Is ButterClaw free?
- ButterClaw offers a free option. Hosting, model usage, or optional services may have separate costs; check the pricing details and official terms.
- Is ButterClaw open source?
- Yes. ButterClaw is open source.
- Does ButterClaw have an API?
- Yes. ButterClaw exposes a developer API. See the official documentation at https://butterclaw.tech for details.
- Can I self-host ButterClaw?
- Yes. ButterClaw supports self-hosting on your own infrastructure.
- What platforms does ButterClaw support?
- ButterClaw is available on: Docker, Linux.
Curated lists that include this category
Authorized agents drift after review
Agents that pass initial checks still pull credentials they should not touch or leak context through indirect prompts. Most observability tools miss these post-authorization changes entirely.
How ButterClaw responds
ButterClaw runs as a self-hosted monitor that watches agent behavior for prompt injection attempts, credential exfiltration, and policy violations. On detection it can revoke OAuth tokens and shred local credentials automatically. Everything executes locally with no cloud telemetry or outbound data.
Key limits and controls
The tool ships under Apache 2.0 with Docker deployment. Operators set the security level from passive observation to active token revocation. Policy tuning and all scaling work fall to the operator with no vendor support path.
Integrations and requirements
It works with Nginx, Gemma 4, ntfy, and SQLite on Docker and Linux. An API is available and the tool is fully self-hosted.
Who it is for / who should skip it
ButterClaw fits self-hosted agent deployments that handle sensitive credentials and need zero-trust checks after authorization. Teams without dedicated security engineering should skip it because policy errors and infrastructure maintenance stay entirely on the operator.
Contribute
- Work on ButterClaw? Claim this listing.
- Embed a badge for your README or site.
Sign in to rate this tool or suggest an edit.