8 comments

  • radku28 minutes ago
    Nice work shipping this.<p>Disclosure: author of a related tool here. I have create agent-vault-proxy for a very similar reason. It also can help keep credentials out of the agent process. The agent gets a placeholder, the proxy swaps in the real secret in transit.<p>I read them as complementary: action firewall in front, credential broker behind. <a href="https:&#x2F;&#x2F;github.com&#x2F;inflightsec&#x2F;agent-vault-proxy" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;inflightsec&#x2F;agent-vault-proxy</a>
  • varmabudharaju50 minutes ago
    This is very interesting. I build something like this but native to claude code and something that focus on just logging the violation. My question is if you are terminating a process with in the workflow will that about all other things that executed before. anyway would love your feed back on this <a href="https:&#x2F;&#x2F;github.com&#x2F;varmabudharaju&#x2F;agent-pd" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;varmabudharaju&#x2F;agent-pd</a>
    • undefined_void25 minutes ago
      claw patrol runs on the network level. There’s no process being terminated - HTTP&#x2F;SQL&#x2F;etc are rejected based on rules that you define. it’s resilient to the agent making changes to its own hooks or bypassing a local sandbox.
    • varmabudharaju50 minutes ago
      *abort
  • Apylon7772 days ago
    This is a really cool library to look at even if you aren&#x27;t running openclaw directly.<p>Lots of good concepts to seek inspiration from.<p>1. process-scoped egress policy<p>2. policy-as-code<p>3. explicit approval classes<p>4. normalized network&#x2F; guardrail receipts.<p>5. structured guardrail outcomes<p>6. centralized decision rules
    • rough-sea1 day ago
      Thanks! Don&#x27;t forget wire level protocol parsing - this is important because agents usually can spawn subprocesses and if they have postgres credentials, you&#x27;re just one psql call away from disaster if you only have MCP&#x2F;HTTP proxies in place.
  • Jayakumark1 hour ago
    How will credentials be injected via Gateway for each user ? If we have 5 users with one gateway, how it knows whose github credential to inject ?
    • rough-sea1 hour ago
      You can define different profiles that are associated with different credentials. Take a look here <a href="https:&#x2F;&#x2F;clawpatrol.dev&#x2F;docs&#x2F;credentials&#x2F;#single-credential-the-common-case" rel="nofollow">https:&#x2F;&#x2F;clawpatrol.dev&#x2F;docs&#x2F;credentials&#x2F;#single-credential-t...</a>
  • dhavd22 minutes ago
    I did this
  • pavelpilyak2 days ago
    Neat! Reading the docs - it&#x27;s default-allow and ships with no rules? Any plans for a default rule set?
    • rough-sea1 day ago
      Yes default allow and no rules by default. Some sort of default policy would be a great feature - I&#x27;ve been considering it. No one wants agents to DROP tables.<p>We have a big and detailed config file for our own internal use - but reluctant to release that exactly because it has information about our systems.<p>There&#x27;s an example config file here that might be helpful <a href="https:&#x2F;&#x2F;github.com&#x2F;denoland&#x2F;clawpatrol&#x2F;blob&#x2F;main&#x2F;examples&#x2F;gateway.example.hcl" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;denoland&#x2F;clawpatrol&#x2F;blob&#x2F;main&#x2F;examples&#x2F;ga...</a> - we use agents to write the config by pointing it at <a href="https:&#x2F;&#x2F;clawpatrol.dev&#x2F;llms-full.txt" rel="nofollow">https:&#x2F;&#x2F;clawpatrol.dev&#x2F;llms-full.txt</a>
  • Hans_Cui5 hours ago
    really interesting work! i am curious how you handle rule configuration for different protocols such as Postgres or ssh. Thanks for open-sourcing it under MIT.
    • rough-sea1 hour ago
      There&#x27;s a plugin API <a href="https:&#x2F;&#x2F;clawpatrol.dev&#x2F;docs&#x2F;plugins&#x2F;" rel="nofollow">https:&#x2F;&#x2F;clawpatrol.dev&#x2F;docs&#x2F;plugins&#x2F;</a>
  • oneclickclaw5 hours ago
    [flagged]