The approval and accountability
layer for agentic AI

Your agents need identity, policy, and human oversight.Not another IAM — the governance layer that's missing.

or try an interactive demo — no signup needed
npm install @sidclaw/sdk
5 lines to govern any tool
// Before: your agent calls tools directly
await sendEmail(customer, subject, body);
// After: one wrapper, full governance
const governed = withGovernance(client, {
operation: 'send_email',
data_classification: 'confidential',
}, sendEmail);
await governed(customer, subject, body);
// → Policy evaluates → Approval if needed → Trace recorded

Your agents are acting without oversight

73%

of CISOs fear agent risks

Only 30% say they are ready

79%

have blind spots in agent operations

37%

already had agent-caused incidents

1 NeuralTrust State of AI Agent Security 2026

How it works

Identity → Policy → Approval → Trace

Identity

Every agent is governed with an owner and scoped permissions.

Policy

Every action is evaluated against explicit rules.

Differentiator

Approval

High-risk actions get human review with rich context.

Trace

Every decision creates an auditable trace.

See exactly what your agent wants to do — and decide

Every high-risk action surfaces a context-rich approval card. One click to approve or deny.

Approval Request
HIGHPENDING

send_email email_service

Customer Communications Agent

2m ago
Resource:customer_emailsconfidential

Outbound customer communications require human review before sending to ensure compliance with FINRA communication standards and data handling policies.

Policy: Outbound customer email reviewVersion: v1

Context Snapshot

{
  "recipient": "[email protected]",
  "subject": "Q4 Portfolio Rebalancing Summary",
  "reason": "Quarterly client portfolio update"
}

What makes this different

CapabilityTraditional IAMSidClaw
Agent Identity
Policy Evaluation
Audit Trail
Approval Workflow← the gap
Context Cards
Risk Classification
Integrity Hashes

Built for regulated industries

Finance

FINRA 2026

Pre-approval of AI use cases, human-in-the-loop for sensitive operations, complete audit trails for regulatory examination.

Healthcare

HIPAA

Scoped agent access to PHI, minimum necessary standard enforcement, approval workflows for data access requests.

Platform Teams

Scale

Govern agents at scale across your organization. Centralized policy management, team-level approval routing, cross-agent audit.

OpenClaw Skills

329K+ agents

OpenClaw has 329K+ stars and 5,700+ skills — but 1,184 malicious skills were found in the ClawHavoc campaign. SidClaw adds the missing policy and approval layer to any OpenClaw skill.

Learn more →

Built for compliance frameworks

FINRA 2026
Agent governance requirements for financial services
EU AI Act
Articles 9, 12, 13, 14 — human oversight and logging
FINMA
Swiss financial regulatory compliance for AI agents
NIST AI RMF
Risk management framework for AI systems
OWASP Agentic
Top 10 for Agentic Applications security

SidClaw maps to FINRA 2026, EU AI Act Articles 9/12/13/14, FINMA operational risk circulars, NIST AI RMF, and the OWASP Top 10 for Agentic Applications.

See it in action

Pick a scenario. Each demo uses real SidClaw governance — only the business data is simulated.

💬
Financial Services

Atlas Financial

AI customer support agent sends emails, looks up accounts, and handles sensitive data. See FINRA-compliant approval workflows in action.

Chat with AI agent
Email approval flow
PII export blocked
Try Demo
FINRA 2026 compliant
📊
DevOps & Platform

Nexus Labs

AI ops agent monitors infrastructure, scales services, and deploys to production. See how governance prevents destructive actions.

Live service monitoring
Production deploy approval
Namespace deletion blocked
Try Demo
Deploy safety controls
🏥
Healthcare

MedAssist Health

AI clinical assistant reviews patient charts and recommends treatments. See HIPAA-compliant controls that keep physicians in the loop.

Patient chart review
Lab order approval
Prescriptions blocked for AI
Try Demo
HIPAA compliant

2 minutes · No signup required · Real governance

Pricing

Free

CHF 0/month

  • 5 agents
  • 10 policies per agent
  • 2 API keys
  • 7-day trace retention
  • 1 webhook
  • Community support
Get Started

No credit card required

Starter

CHF 199/month

Cancel anytime

  • 15 agents
  • 50 policies per agent
  • 5 API keys
  • 30-day retention
  • 3 webhooks
  • Email support
Start Starter
Most Popular

Business

CHF 999/month

For production teams

  • 100 agents
  • Unlimited policies
  • 20 API keys
  • 90-day retention
  • 10 webhooks
  • SSO/OIDC
  • Priority email support
Start Business

Enterprise

From CHF 3,000/mo

Self-hosted or cloud

  • Unlimited agents
  • Unlimited policies
  • Unlimited API keys
  • Custom retention
  • Unlimited webhooks
  • Self-hosted in your VPC
  • Dedicated support & SLA
  • Compliance documentation
  • FINMA/EU AI Act mapping
Contact Sales

Founding Customer Offer

First 10 customers get 50% off the first year. Contact [email protected] to claim your spot.

Deploy anywhere

The SDK is Apache 2.0 — use it anywhere, no restrictions. The platform is source-available under the Functional Source License (FSL) — inspect every line, audit it yourself. After two years, all code converts to Apache 2.0.

Hosted cloud

We run it, you use it. Start free, scale as you grow. Zero infrastructure to manage.

Self-hosted

Deploy in your own VPC, on-premises, or air-gapped environment. Enterprise license includes support, SLA, and compliance documentation.

Your governance data never leaves your infrastructure unless you choose our cloud.

The governance SDK will always be free and open. We offer hosted convenience for teams that want it, and enterprise licenses with support for organizations that self-host.

Get started in 2 minutes

or

npm install @sidclaw/sdk