Let your AI agents run fast without tripping
Terminal-based permissions only work when you're watching. Bulwark gives you a centralized dashboard — approve from your phone, block bypass flags, see every session.
Pending Approvals
3 waitingrm -rf ./tests/integration/Session: "refactor auth module"
Terminal permissions are a good start. But they're not enough.
Built-in permission systems work great for interactive use. But when agents run in CI/CD, background tasks, or across a team — you need more.
Stuck in the Terminal
Built-in agent permissions only work when you're watching the session. Step away and your agent waits. Run it in the background? No approval possible.
Easily Bypassed
CLI flags like --dangerously-skip-all-permissions exist. One flag and all your carefully configured rules are ignored. No audit trail, no enforcement.
No Visibility Across Sessions
Running multiple agents? Each one is isolated. No centralized view of what's happening, no single place to manage approvals across your team.
Bulwark extends Claude Code with centralized, enforceable, bypassable-proof permissions that work everywhere your agents run.
Control your agents, not your terminal
Bulwark extends your agent's permissions into a centralized system that works even when you're not watching.
Install & Connect
One CLI command hooks Bulwark into your coding agent. All tool calls are routed through your centralized policy engine.
Set Your Policies
Define what gets auto-approved, what gets blocked, and what needs your sign-off. Policies can't be bypassed with flags.
Approve From Anywhere
Pending requests appear in your dashboard. Approve from your phone, laptop, or anywhere — no terminal required.
What Bulwark adds to your agents
Keep the permissions you know, add the control you need for real-world agent deployments.
Approve From Anywhere
Dashboard accessible from any device. Approve requests from your phone while the agent runs on your workstation.
Bypass-Proof
--dangerously-skip-all-permissions? Doesn't work. Bulwark enforces policies at the network level, not in the CLI.
Background Agent Ready
Agents in CI/CD, cron jobs, or headless mode can request approvals. No terminal session required.
Centralized Policies
One policy set for your entire team. Consistent rules across all sessions, all devices, all agents.
Full Audit Trail
Every tool call logged with context. See what your agents tried to do, what was approved, and by whom.
Smart Auto-Approve
Not everything needs review. Define patterns for safe operations and let them fly through automatically.
Works with your favorite coding agents
Seamless integration with the tools you already use. More integrations coming soon.
More integrations coming soon
Get started in seconds
Three commands and you're protected. No config files, no complex setup.