Skip to content
Interactive Product Demo

Experience the platform.

See how Curate-Me governs your OpenClaw agents in real time. No signup required — explore every feature below.

migration.sh

The safety pipeline.

Every AI request passes through 5 policy gates. Short-circuits on first denial. Click below to watch it run.

Incoming Request

POST /v1/chat/completions · model: claude-opus-4 · 1,247 tokens

acme-corp

Rate Limit

Per-org request throttling

Cost Estimate

Budget check before execution

Personal Data Scan

Regex scan for secrets & PII

Model Allowlist

Enforce permitted models

Human Approval Gate

Human approval for sensitive ops

Runner lifecycle.

From setup to completion — every state change is recorded, auditable, and controllable.

Idle

Runner template selected, awaiting deployment command.

State 1/6
$ curate runners create --template code-review --sandbox WRITE_PROJECT

41 pages of visibility.

Runners, costs, security, compliance, audit trails, templates, schedules, credentials — all in one console.

dashboard.curate-me.ai/runners
8+2
Active Runners
$4.82-12%
Today's Cost
98%+3
Security Score
42
Requests/min
Request Volume (24h)1,847 total

Three SDKs. One API.

Python, TypeScript, or CLI — full runner lifecycle management from any environment.

from curate_me import CurateMe

client = CurateMe(api_key="cm_sk_live_...")

# Create a governed runner
runner = client.runners.create(
    template="code-review",
    sandbox_tier="WRITE_PROJECT",
    daily_budget=25.00
)

# Start a session
session = client.runners.start_session(
    runner_id=runner.id,
    task="Review PR #142 for security issues"
)

# Stream logs in real time
for event in client.runners.stream_logs(runner.id):
    print(f"[{event.type}] {event.message}")

# Get cost summary
costs = client.runners.get_costs(runner.id)
print(f"Total: {costs.total}")  # $2.14
$pip install curate-me

Built for production.

65+ backend modules. 24+ dashboard pages. Every OpenClaw pain point, addressed.

Access Tier Isolation

4-tier access control. Deny patterns block .env, .pem, .git/config.

Network Phases

SETUP (on) → EXECUTION (off) → TEARDOWN (on). No exfiltration.

Cost Governance

Per-request limits, daily budgets, auto-terminate on overspend.

Human Approvals

Approval queues for high-cost or sensitive agent operations.

Personal Data Scanning

Block API keys, passwords, and personal data before they hit any LLM.

Desktop Streaming

Real-time VNC. Watch your agent work. Take over anytime.

Git Worktrees

Per-session branch isolation. No cross-contamination.

Audit Trail

27 event types. Immutable log. Queryable and exportable.

Ready to govern your agents?

Stop patching CVEs. Stop praying your agent doesn't overspend. Start shipping with confidence.

No credit card 14-day free trial Migrate in 5 minutes