Rulebricks Guardrails

We have since built a dedicated product to help developers manage AI permissions better, which we intend to make open source. Check it out at stoplight.ai.

Claude Code Guardrails

settings.json works if:

- You're fine editing JSON and killing Claude Code sessions every time policy changes
- Your security team is comfortable making PRs
- You don't need to know what got blocked, when, or for whom
- Basic pattern matching like Bash(rm:*) covers your use cases

Use this if:

- Policy changes need to apply instantly across your team—no git pull, no restart
- Security/compliance needs a clear audit trail of every blocked command
- You need conditional logic: "allow rm -rf on node_modules, deny everywhere else"
- Non-engineers need to edit rules without touching config files

Rulebricks gives you instant governance from one hook.

Setup (5 minutes)

1. Create your rules

1. Go to rulebricks.com and create an account
2. Fork one of these templates from the "AI Agents" category:
   - Claude – Bash Guardrails — control shell commands
   - Claude – File Access Policy — control file read/write/edit
   - MCP Tool Governance — control MCP server operations
3. Customize the rules for your team
4. Publish the rule
5. Copy your API key from the API tab

2. Install

Automatically finds your rules

Claude will detect your published rules and wire up the appropriate hooks.

3. Restart Claude Code

You're done.

What gets checked

Configuration

Environment variables in ~/.claude/settings.json:

Updating rules

Edit your decision table and publish a new version. Changes apply immediately— no restart, no redeployment.

Reviewing histories

Review the history of blocked commands in the Logs tab. You can query by tool, approval decision, and more. There are other meaningful perks to this…