Clawsec
Clawsec is an open-source, YAML-configurable security middleware for OpenClaw agents that intercepts every tool invocation and runs it through a rules engine in under 5ms.
About
Clawsec is an open-source, YAML-configurable security middleware for OpenClaw agents that intercepts every tool invocation and runs it through a rules engine in under 5ms. Built-in rulesets cover destructive commands (rm -rf, DROP TABLE), secret exposure, data exfiltration, unauthorized purchases, privilege escalation, and prompt injection. Each rule can be configured to block, require human confirmation, or allow-with-log. One-command install via OpenClaw plugins.
OpenClaw developers who want a practical, low-friction security layer with configurable rules for blocking or gating dangerous agent actions — particularly teams who've been running agents without guardrails and want a quick path to meaningful runtime security.
Pros & Cons
Pros
- check Three-action model (block, confirm, allow+log) gives fine-grained control rather than all-or-nothing blocking
- check Under 5ms rule evaluation latency means minimal impact on agent execution speed
- check YAML-extensible rulesets: extend built-in profiles (aws-security, pii-protection) or write custom rules
- check Covers six distinct threat categories out of the box — one of the more comprehensive rule sets in open-source agent security
- check MIT licensed and open source with community-contributed rules shipping regularly
Cons
- close OpenClaw-specific — not a generic agent security framework; other runtimes need adaptation
- close Pattern matching is inherently incomplete; sophisticated prompt injection or obfuscated commands can bypass rule-based detection
- close Human confirmation workflows (agentConfirm) add friction in high-volume autonomous agent pipelines
- close No hosted/managed option — teams must maintain their own clawsec.yaml and rule updates
More Security
Other tools in the same category.
