# denoland/clawpatrol

Security firewall for agents

Repository: https://github.com/denoland/clawpatrol
Canonical: https://ross.abutalabs.com/products/clawpatrol
Homepage: https://clawpatrol.dev
Language: Go
License: MIT
License Family: permissive
Last push: 2026-08-19T03:28:21+00:00

## Health v2 (maintenance only)
Score: 80/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 98, longevity 9
- inputs: {"age_days": 127, "days_push": 14, "days_rel": 14, "gap_med": 0, "n_releases_24m": 30}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1030, forks 52 (observed 2026-08-28T04:03:17.706109+00:00)

## What it is
Claw Patrol is a security firewall and proxy for AI agents, written in Go, that intercepts agent traffic at the wire level and gates each action against rules written in HCL with CEL conditions. It supports HTTP, SQL (Postgres/ClickHouse), Kubernetes, and SSH protocols, with human or LLM approvers, credential safeguarding, audit logging, and an operator dashboard.

## Use cases
- block destructive SQL from AI agents
- require human approval for kubectl delete commands
- prevent agent credentials from leaking via prompt injection
- audit log everything an AI agent does
- gate agent HTTP requests through an LLM judge
- sandbox one agent process's network traffic
- enforce access policies for agents talking to Kubernetes

## When to choose
- you run autonomous AI agents with access to production databases, Kubernetes, or APIs and need policy enforcement
- you want human-in-the-loop approval for risky agent actions
- you need a single self-hosted gateway with audit logs and credential protection for multiple agent machines

## When to avoid
- you only need simple API rate limiting or a traditional web API gateway without agent-awareness
- your agents run on Windows or other unsupported platforms (Linux/macOS amd64/arm64 only)
- you need a fully managed SaaS rather than a self-hosted gateway you operate

## Facets
- artifact type: service
- maturity: active
- function: security, proxy, api-gateway, middleware, secrets-management, monitoring, logging, plugin-system, configuration-management
- domain: security, developer-tools, self-hosted, networking
- platform: go, self-hosted, cli
- tags: agent-firewall, ai-agent-security, policy-engine, cel-expressions, hcl-config, wireguard, prompt-injection, audit-log, human-in-the-loop, llm-judge, traffic-interception, tailscale, ai-agents, devops, linux, macos, docker

## Member repositories
- denoland/clawpatrol (main) score 80

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:17.706109+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:07:32.916967+00:00, confidence not recorded.
  - readme: https://github.com/denoland/clawpatrol (fetched 2026-08-28T04:03:17.706109+00:00, sha 1fb1c6f52dcf)
  - homepage: https://clawpatrol.dev (fetched 2026-08-29T13:07:03.264944+00:00, sha 254c00cae42c)
  - site_page: https://clawpatrol.dev/docs/getting-started (fetched 2026-08-29T13:07:03.276281+00:00, sha b39247ac7968)
  - site_page: https://clawpatrol.dev/docs/rules (fetched 2026-08-29T13:07:03.278036+00:00, sha c8a244f29d13)
  - site_page: https://clawpatrol.dev/docs/plugins (fetched 2026-08-29T13:07:03.280843+00:00, sha c4c5b20d82f0)
  - site_page: https://clawpatrol.dev/docs (fetched 2026-08-29T13:07:03.274604+00:00, sha 96c6ab6907e7)
- Data as of 2026-08-30T08:39:29.467469+00:00.