SidClaw GovernanceMCPServer

Your agents call tools without oversight. SidClaw intercepts every tool call, checks it against your policies, and holds risky actions for human review before they execute.

Opens a local governance dashboard athttp://localhost:3030with four pre-loaded scenarios (Claude Coderm -rf, fintech trade, DevOps scale-to-zero, clinical lab order). No signup, no Docker, no API key — just the approval card UX running in your browser.

Agent wants to send an email → policy flags it → reviewer sees full context → approves or denies → trace recorded.

Customer Support Agent (Financial Services)

An AI agent wants to send a customer email. Policy flags it for review. The reviewer sees full context — who, what, why — and approves with one click. Every step is traced.

An AI agent wants to scale production services. High-risk deployments require human approval. Read-only monitoring is allowed instantly.

An AI assistant recommends lab orders. The physician reviews the clinical context and approves. Medication prescribing is blocked by policy — only physicians can prescribe.

Agent wants to act → SidClaw evaluates → Policy decides → Human approves (if needed) → Action executes → Trace recorded

Four primitives govern every agent action:

┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ Identity │ → │ Policy │ → │ Approval │ → │ Trace │ │ │ │ │ │ │ │ │ │ Every │ │ Every │ │ High-risk│ │ Every │ │ agent │ │ action │ │ actions │ │ decision │ │ has an │ │ evaluated│ │ get human│ │ creates │ │ owner & │ │ against │ │ review │ │ tamper- │ │ scoped │ │ explicit │ │ with rich│ │ evident │ │ perms │ │ rules │ │ context │ │ audit │ └──────────┘ └──────────┘ └──────────┘ └──────────┘

- allow→ action executes immediately, trace recorded
- approval_required→ human sees context card, approves/denies, trace recorded
- deny→ blocked before execution, no data accessed, trace recorded

Railway is the recommended one-click deploy — it spins up Postgres + API + Dashboard together. Vercel hosts only the Next.js dashboard; pair it with a hosted API.

https://vercel.com/new/clone?repository-url=https%3A%2F%2Fgithub.com%2Fsidclawhq%2Fplatform&root-directory=apps%2Fdashboard&env=NEXT_PUBLIC_API_URL&envDescription=Your%20SidClaw%20API%20base%20URL%20(e.g.%20https%3A%2F%2Fapi.sidclaw.com)

Vercel can only host the dashboard (Next.js). The API is Fastify — deploy it to Railway, Fly, Render, or run via Docker. SetNEXT_PUBLIC_API_URLon the dashboard project to point at it.

Under 3 minutes to a working instance on Railway.

For Claude Code users. EveryBash,Write,Agent,mcp__tool call is governed by SidClaw:

# In the SidClaw platform repo npm run hooks:install # Then set two env vars export SIDCLAW_BASE_URL=https://api.sidclaw.com export SIDCLAW_API_KEY=ai_your_key_here

AI agents are being deployed in production, but the governance layer is missing:

The big vendors (Okta, SailPoint, WorkOS) handle identity and authorization. But none of them ship theapproval step— the part where a human sees rich context and makes an informed decision before an agent acts.

SidClaw maps to regulatory requirements across the US, EU, Switzerland, and Singapore:

- 60-second setup—npx create-sidclaw-appscaffolds a working governed agent
- <50ms evaluation overhead— the governance layer is invisible to your users
- 5-minute integration— wrap existing tools, no code changes
- MCP-native— governance proxy for any MCP server
- Framework-agnostic— LangChain, Vercel AI, OpenAI, CrewAI, Pydantic AI, Composio, Claude Agent SDK, Google ADK, LlamaIndex, NemoClaw, or plain functions
- Typed SDKs— TypeScript (npm) + Python (PyPI)

- Policy engine— allow / approval_required / deny with priority ordering and classification hierarchy
- Approval workflow— context-rich cards with agent reasoning, risk classification, and separation of duties
- Audit trails— correlated traces with integrity hash chains (tamper-evident)
- SIEM export— JSON and CSV, continuous webhook delivery

- RBAC— admin, reviewer, viewer roles with enforced permissions
- Tenant isolation— automatic tenant scoping on every query
- API key management— scoped keys with rotation
- Rate limiting— per-tenant, per-endpoint-category
- Webhooks— real-time notifications for approvals, traces, lifecycle events
- Chat integrations— approve/deny from Slack, Teams, or Telegram without opening the dashboard
- Self-serve signup— GitHub, Google, email/password

┌─────────────┐ ┌──────────────┐ ┌──────────────────┐ │ Your Agent │ │ SidClaw SDK │ │ SidClaw API │ │ │ ──► │ │ ──► │ │ │ LangChain │ │ evaluate() │ │ Policy Engine │ │ MCP Server │ │ withGovern() │ │ Approval Service │ │ OpenAI SDK │ │ governTools()│ │ Trace Store │ │ Any tool │ │ │ │ Webhook Delivery │ └─────────────┘ └──────────────┘ └──────────────────┘ │ ┌────────┴────────┐ ▼ ▼ ┌──────────────┐ ┌──────────────┐ │ Dashboard │ │ Notifications│ │ │ │ │ │ Agents │ │ Slack │ │ Policies │ │ Teams │ │ Approvals │ │ Telegram │ │ Traces │ │ Email │ │ Settings │ │ Webhooks │ └──────────────┘ └──────────────┘

Deploy from the GitHub repo to Railway. Add a PostgreSQL database, configure environment variables, and you're live.

Deploy the dashboard to Vercel (requires a separately hosted API).

curl -sSL https://raw.githubusercontent.com/sidclawhq/platform/main/deploy/self-host/setup.sh | bash

git clone https://github.com/sidclawhq/platform.git cd platform cp deployment/env.example .env # edit with your values docker compose -f docker-compose.production.yml up -d

The SDK (packages/sdk/) is Apache 2.0. The platform (apps/) is FSL 1.1.

This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.

Require a named human's offline-verifiable approval before an AI agent takes an irreversible action — payment release, record change, deploy. Two-person rule, Ed25519 Trust Receipts, IETF-drafted, Apache-2.0.

Wrap your local MCP servers to check every tool call against policy and live state before it runs - allow, block, or request a refresh, with a reason the agent can act on.

Paid remote MCP for AI agent safety replay checks, policy gates, eval receipts, control-fix suggestions, and release evidence exports.

Connect to your CISO Adapt workspace to search, analyse, export, and manage risks and policies with natural language

Boost security in your dev lifecycle via SAST, SCA, Secrets & IaC scanning with Cycode.

Cloud security insights, guardrail guidance, and compliance checking via Dawnguard.

Six-gate governance for AI agents: PROCEED/PAUSE/HALT decisions with hash-chained audit trails.

Paid remote MCP for agent data-access boundary reviews, permission scope evidence, sensitive data notes, and governance receipts.

Remote MCP server (Streamable HTTP) at https://mcp.agenticrail.nz/ — deterministic step-order enforcement for AI agents. evaluate_step returns ALLOW or DENY before a step runs; verify_receipt proves a sequence's Ed25519-signed, hash-chained receipt chain is intact. No auth required: omit the bearer token and calls run on the public demo key. That first clause matters — the form has no "remote/hosted" field, and putting the endpoint in the description is the convention on that list ("Fully REMOTE! Just use…"). The rest mirrors your own server card verbatim, so the listing and the card can't drift.