Product

BlackLake Surface and Depth

Two products that solve complementary problems. Surface gives you visibility and control over every agent tool call. Depth gives your agent workflows durable execution with crash recovery. Use one or both — they work better together.

npx @blacklake-systems/surface-cli

npx @blacklake-systems/depth-cli run workflow.ts

Getting started Read the docs

Surface

Visibility and control for AI agents

Surface runs on your machine and sits between your AI tool and the services your agents connect to. MCP (Model Context Protocol) is the open standard AI tools use to connect to external services. Surface proxies those connections: no integration code required, policy engine for rules, approval workflows for human oversight, and cost tracking across all providers.

MCP Client

Your AI Tool

Surface

PolicyApprovalsAuditCost

MCP Servers

Your MCP Servers

Surface capabilities

What you get

MCP Proxy

Sits between your AI tool and the services your agents connect to. Every tool call flows through BlackLake. No integration code required: configure and go.

Policy Engine

Declarative rules that gate what agents can do, enforced at the tool-call layer before actions execute. Priority-ordered, selector-based.

Approval Workflows

Route sensitive actions to human reviewers with full context, then record the decision. The MCP proxy holds the call open until you decide.

Cost Tracking

Proxy your LLM API calls through BlackLake. See token counts, model usage, and costs across Anthropic, OpenAI, Ollama, and others, by day, week, or month.

Audit Log

Every tool call, every decision, every approval. Immutable, queryable, and all in one place.

One Command

No Docker. No database setup. No signup. npx @blacklake-systems/surface-cli starts everything on your laptop.

Surface integration

Two ways to connect

Use the MCP proxy (no integration code required) or the SDK for deep integration into your own agents. Or both.

MCP Proxy

No integration code required.

Point your MCP servers through BlackLake. Every tool call is evaluated, logged, and visible in the dashboard. Set “allow”, “deny”, or “ask” per server.

MCP configuration

// ~/.blacklake/mcp-config.json{  "servers": {    "filesystem": {      "command": "npx",      "args": ["-y", "@modelcontextprotocol/server-filesystem", "."],      "policy": "ask"    }  }}

TypeScript SDK

For anyone integrating oversight into agent code.

Integrate oversight directly into your agent code. Call bl.govern() before any tool invocation for policy evaluation, approval routing, and audit logging.

SDK integration

import { BlackLake } from '@blacklake-systems/surface-sdk';const bl = new BlackLake({  apiKey: process.env.BLACKLAKE_API_KEY,  baseUrl: 'http://localhost:3100',});const decision = await bl.govern({  agent: 'expense-bot',  tool: 'stripe.refund',  action: { amount: 4200 },});if (decision.decision === 'allow') {  await stripe.refunds.create({ amount: 4200 });}

npx @blacklake-systems/surface-cli

Depth

Durable execution for agent workflows

Write AI workflows as TypeScript async functions. Each step persists to disk as it completes. If the process crashes, re-run the file — completed steps replay from SQLite instantly and execution resumes from where it stopped. Works standalone or with Surface for governed, audited workflows.

Step-based execution

Write workflows as TypeScript async functions. Each step persists to disk.

Crash recovery

If the process dies, re-run the file. Completed steps replay from SQLite instantly.

LLM routing

Call Anthropic, OpenAI, or Ollama with one API. Bring your own credentials.

Surface integration

Tool calls go through Surface's governance. Costs show in Surface's dashboard.

Typed errors

ToolDeniedError, ToolNotFoundError, SurfaceUnavailableError — not generic catches.

Local executors

Depth workflow

import { workflow, step } from '@blacklake-systems/depth-sdk';export default workflow('research', async (ctx) => {  const data = await step(ctx, 'gather', async () => {    return await ctx.llm('anthropic:claude-sonnet-4-6', {      prompt: 'Find recent papers on AI governance',    });  });  await step(ctx, 'save', async () => {    await ctx.tool('filesystem.writeFile', {      path: './report.md', content: data,    });  });});

npx @blacklake-systems/depth-cli run workflow.ts

Better together

Surface and Depth, integrated

When both products are running, they connect automatically. Tool calls made inside Depth workflows route through Surface’s governance engine. Costs from both products appear in a single dashboard. Approval requests block the Depth workflow until a human decides.

Governed tool calls

Every tool call a Depth workflow makes is evaluated against Surface policies before it executes.

Approvals in the console

Sensitive steps pause the workflow and surface an approval request in the Surface console. Approve or reject — the workflow continues or aborts.

Unified cost tracking

LLM spend from Depth workflows is attributed and shown alongside direct Surface usage in one cost view.

Deployment

Local first. Cloud when you need it.

Local

Free, forever.

Runs on your laptop with SQLite
Surface dashboard at localhost:3200
MCP proxy, policy engine, approvals, cost tracking
Depth workflows with full crash recovery
No cloud account, no Docker, no signup

Cloud

When you need more.

Access from any device, including your phone
Persistent history that survives machine restarts
Team visibility across multiple developers
Sync local data to the cloud console

Get started in one command

No signup. No credit card. Runs on your machine.

npx @blacklake-systems/surface-cli

npx @blacklake-systems/depth-cli run workflow.ts

Getting started guide Read the docs