Veto Documentation

Veto is a permission gateway for AI coding agents. It sits between your AI tools (like Claude Code) and the actions they perform, giving you centralized control over what's allowed, denied, or flagged for human review.

What Veto does

  • Intercepts tool calls — every file write, bash command, or API call your AI agent makes passes through Veto first
  • Evaluates against your rules — pattern-based rules (regex) decide instantly whether to allow or deny
  • AI-powered scoring — when no rule matches, an AI model evaluates the risk of the action
  • Logs everything — full audit trail of every tool call and decision, searchable from the dashboard

Two integration modes

Claude Code Plugin

For direct Claude Code usage. Install the Veto plugin from the marketplace (Linux/macOS or Windows) — it intercepts PermissionRequest events before they execute.

Claude Code → Veto Plugin (hook) → Veto Server → allow/deny

LLM Proxy

For team-wide enforcement. Veto includes a built-in LLM proxy that sits between Claude Code and the Anthropic API. Enable it from the dashboard — tool calls in the LLM response are evaluated before reaching the client.

Claude Code → Veto LLM Proxy → Anthropic API
                    ↓
              Veto Server → allow/deny

Next steps