$ loading_
Run autonomous AI agents securely with plain-English policy constraints.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "ironcurtain" yet — see the docs or source repo.
Use ironcurtain to create a runtime policy for an autonomous AI agent that can access websites and local files. Define plain-English rules that forbid file deletion, block access to directories containing secrets, allow reading docs/ and src/, and allow only whitelisted domains. Provide an executable policy example.
A clear plain-English safety policy example specifying allowed and forbidden actions for runtime enforcement.
I have an autonomous AI agent with shell, web request, and database query capabilities. Using ironcurtain, design runtime restrictions based on least privilege: restrict dangerous shell commands, limit outbound network access, and make the database read-only. Explain what risk each rule reduces.
A least-privilege runtime policy with explanations of the risks mitigated by each restriction.
Draft an ironcurtain-style constitutional policy for a production customer support automation agent. Include: protecting user privacy, preventing unauthorized actions, escalating high-risk requests to humans, and logging key decision rationales. Output a structured list of clauses.
A structured constitutional policy that serves as high-level safety and behavior governance for a production agent.
Add human approval and tamper-evident logs to risky AI agent actions.
Control autonomous AI agents with policies, approval gates, and audit trails.
Run AI coding agents safely in a self-hosted, isolated sandbox environment.
Run AI agents with fine-grained access control and zero-setup brokering.
Intercept and block MCP tool calls with YAML policies for safer AI agents.
Gives AI agents runtime perception to understand web applications they build.