Self-host

Run Clawvisor on your own infrastructure.

Full control over your data and configuration. The gateway is the same one that powers the managed cloud — you operate it, and we never see your traffic.

Run Clawvisor on a separate host from your agent

Install Clawvisor on a distinct computer, container, or VM from the agent it gates. Co-locating them lets a compromised or misbehaving agent bypass approval flows by reading credentials or tampering with the daemon directly — the isolation boundary is what makes the security model work.

Step 1

Install the daemon.

The installer drops the daemon into ~/.clawvisor/bin. The GitHub repo is the canonical setup guide — it covers configuration, the datastore, and connecting your accounts.

install
$ curl -fsSL https://raw.githubusercontent.com/clawvisor/clawvisor/main/scripts/install.sh | sh

Or build from source. Requires Go 1.25+ and Node.js 18+.

build
$ git clone https://github.com/clawvisor/clawvisor.git$ cd clawvisor$ make install

Step 2

Connect your agent

Add the Clawvisor skill to your agent and point it at the gateway on port 25297. The skill teaches it to fetch data through Clawvisor instead of calling services directly — no SDK, no changes to your agent’s code. The repo has the per-harness steps.

Step 3

Give it tasks

When your agent needs your data, you approve the task once and Clawvisor handles the rest — verifying each request against the purpose you approved, injecting credentials from the vault, and writing every decision to the audit log.

  • Requests that match the approved purpose run straight through
  • Off-purpose requests are rejected and logged
  • Injected instructions never reach your accounts

Same gateway, hosted.

Clawvisor Cloud is the same gateway, hosted, with risk scoring and longer log retention. Free to start.

Try Clawvisor CloudRead the source