Know what your AI agents cost, block what they shouldn't do.

Self-hosting OpenClaw means you handle setup, patching, security hardening, and cost monitoring yourself. With Curate-Me managed hosting, we run your OpenClaw instance on dedicated cloud infrastructure with auto-patching for CVEs, real-time cost governance, personal data scanning, and a full ops dashboard -- all included. You get production security without the DevOps burden.

We offer a free tier with 1,000 requests per day and full governance -- no credit card required. The Starter plan is $49/month and includes 100K requests, 3 connected machines, model allowlists, and cost controls. The Growth plan is $199/month with 500K requests, 5 machines, human approval workflows, time-travel debugging, and priority support. Enterprise plans include unlimited requests, self-hosted machines, SSO/SAML, and dedicated infrastructure -- pricing is tailored to your usage. AI provider costs (OpenAI, Anthropic, etc.) are billed directly by the provider -- we never mark up API costs.

OpenClaw has had 6+ CVEs disclosed in a three-week period, including one-click RCE vulnerabilities. Our managed instances receive automatic patches within hours of disclosure. We also run continuous posture assessments, scan all ClawHub skills before installation, and enforce network phase separation so agents cannot make outbound calls during execution. Every action is logged in an immutable audit trail.

Yes. Migration takes under 30 minutes for most setups. We provide a migration CLI tool that exports your existing configuration, skills, and credentials, then imports them into a managed instance. Your agent definitions, prompt templates, and workflow configurations transfer directly. We also offer white-glove migration assistance on Growth and Enterprise plans.

Yes — connect any machine (laptop, server, cloud VM) to Curate-Me in under 2 minutes. Install our lightweight agent, register with a one-time token, and your machine appears in the dashboard with full cost tracking, policy enforcement, and observability. All communication is outbound-only — no ports to open. This is ideal for teams with data residency requirements or existing cloud contracts. Available on Growth and Enterprise plans.

Every request passing through the gateway gets a real-time cost estimate before it reaches the AI provider. You can set per-request cost limits, daily budget caps per organization, and model allowlists to prevent expensive model usage. If a budget threshold is exceeded, the request is denied automatically. An emergency kill switch can halt all agent activity instantly. Cost data feeds real-time dashboards and is persisted for audit.

The ClawHub skills registry has a serious malware problem -- 20% of skills have been flagged as malicious, including credential theft and crypto mining payloads. Curate-Me provides a vetted skills allowlist and a free skill scanner that gives every skill an A-F security grade based on static analysis, dependency audit, and behavior profiling. Only approved skills can be installed on managed instances.

Pro plans include a 99.9% uptime SLA for the managed infrastructure. Enterprise plans offer 99.95% with dedicated support engineers and priority incident response. Our infrastructure runs on dedicated cloud compute with automated health checks, auto-restart on failure, and geographic redundancy options. The governance gateway itself is stateless and horizontally scalable.

Agent execution data, audit logs, and cost records are stored in databases isolated per organization. Personal data scanning runs locally within the gateway before requests leave your network boundary -- sensitive data is never forwarded to AI providers. We do not train models on your data. Enterprise plans support data residency in EU or US regions, and self-hosted machines keep all data on your own infrastructure.

It depends on the model, but most teams spend $50-500/month on API calls. The problem is unpredictability -- a single runaway loop can cost thousands. With Curate-Me, you see exactly what each agent costs per request, per day, and per month. Set a daily budget cap and we block requests before you overspend.

About 5 minutes. You change one environment variable (your base URL) to point at our gateway instead of directly at OpenAI/Anthropic/Google. No code changes, no SDK installs, no infrastructure to manage. Your existing code works unchanged -- we just sit in between and add governance.

Yes. The gateway supports 50+ AI providers including OpenAI, Anthropic (Claude), Google (Gemini), DeepSeek, Groq, Mistral, xAI, and more. You can route different agents to different providers, set model allowlists per organization, and use model aliases to swap providers without changing agent code. The safety pipeline (rate limiting, cost tracking, personal data scanning, human approvals) applies uniformly across all providers.