Is Smoke Monkey an AI model?

No. Smoke Monkey is not an LLM. It is an embeddable, framework-agnostic runtime that turns LLMs (NVIDIA, Gemini, OpenAI, Claude, Ollama) into autonomous engineering agents capable of multi-step planning, calling tools, editing files, and running test loops with zero runtime dependencies.

Do I need a database (Postgres, SQLite, or Redis)?

No database is required. State is persisted through lean session files on disk, or you can plug in your own custom storage adapter. You can deploy it as a CLI, a desktop app, or inside a serverless function without provisioning a database.

How does human-in-the-loop permission gating work?

Smoke Monkey comes with three permission policies: allow-all, deny-all, and ask-default (or a custom function resolver). When a sensitive tool like edit_file or run_command is called under "ask", the runtime emits a permission.required event, pausing execution until your code or UI calls agent.resolvePermission(id, "allow" | "deny").

How does the Model Context Protocol (MCP) server work?

Smoke Monkey ships an official stdio MCP server in smoke-monkey-harness-mcp. Point Claude Code, Cursor, Codex, or opencode at it using npx -y smoke-monkey-harness-mcp to let other AI tools plan, scaffold, verify, and run agents using Smoke Monkey automatically.

What are SKILL.md skills and how do they prevent context bloat?

Skills are directories containing a SKILL.md workflow file. The system prompt only includes a one-line catalog of available skills. When a task requires specialized expertise (e.g. frontend-ui-engineering, test-driven-development, or security-hardening), the agent calls use_skill to load instructions just-in-time, saving thousands of tokens.

What is an AI Agent Harness?

An AI agent harness is the execution environment that bridges raw Large Language Models (LLMs) to real-world software engineering capabilities. While an LLM generates tokens, an AI harness manages deterministic phase loops (explore, plan, edit, verify, recover), tool execution sandboxing, token budget compaction, and safety gates needed for autonomous problem-solving.

What makes Smoke Monkey an Agentic AI runtime rather than a simple wrapper?

Unlike basic prompt-response chains, Smoke Monkey implements true agentic AI architecture: a deterministic 6-phase state machine with automatic failure recovery, runaway loop guards, 24 built-in developer tools, dynamic subcontext isolation, and interactive human-in-the-loop permission pauses.

Is Smoke Monkey Harness 100% Open Source AI?

Yes. Smoke Monkey Harness is 100% open source under the permissive MIT license with zero runtime dependencies. There is no telemetry lock-in, no mandatory cloud backend, and no database requirement—giving engineering teams complete sovereignty over their agent code.

What Agent APIs does Smoke Monkey Harness provide?

Smoke Monkey provides composable TypeScript Agent APIs: createAgent() for initialization, agent.run() for multi-turn execution, agent.on() for fine-grained lifecycle and streaming events, agent.resolvePermission() for interactive human authorization, subcontext switching APIs, and stdio Model Context Protocol (MCP) server endpoints.

Can I run completely offline without API keys?

Yes! Smoke Monkey natively supports Ollama out of the box. Set provider: "ollama" and model: "qwen3:8b" with workspacePath, and the agent will run locally on your machine with zero external network requests.

Autonomous Agent Engine

Smoke Monkey Documentation

Smoke Monkey is the zero-dependency TypeScript runtime designed to help developers build autonomous looping AI agents, code editors, and agentic workflows with React, Next.js, and Node.js.

Why use Smoke Monkey Harness?

Building AI agents with unconstrained ReAct loops or monolithic frameworks results in infinite hallucination loops, context window exhaustion, and opaque failures. Smoke Monkey replaces black-box abstractions with a deterministic 6-phase state machine, automated loop guards, and a clean separation of concerns:

0 Runtime Dependencies

Pure Node.js built-ins. Runs in lightweight workers, microVMs, or servers.

Deterministic 6 Phases

Explore → Plan → Edit → Verify → Recover → Complete.

Human-in-the-Loop Pauses

Non-blocking pauses for tool permissions, questions, and external MCP approvals.

Interactive Architecture & Phase Flow

Pan, zoom, drag nodes, and switch tabs below to inspect how tasks move through the 6-phase cycle, human pauses, and the 3-pillar ecosystem:

Interactive Engine Flow

Drag nodes · Scroll to pan & zoom

🎯 User Prompt "Fix auth bug in auth.ts"

1. EXPLORE Inspect · Glob · Grep

2. PLAN Sub-tasks · JIT Skills

3. EDIT Line-level Patch

4. VERIFY Run Tests & Diff

5. COMPLETE Session Persist

🔁 RECOVER (Demotion Loop) Inspect stderr · Rollback · Auto-heal

🛡️ Automated Loop Guards (Max 1000 Steps · 3x Identical Error Abort · 90% Compaction)

Deterministic loop: verification failure triggers an automated recovery demotion instead of halting.

Live Diagram

The 3 Pillars of Smoke Monkey

Execution is distributed across three modular libraries designed to work in synergy or independently:

Smoke Monkey 3-Pillar Architecture Diagram: React Chat UI, Backend Harness Core Engine, and Model Context Protocol Server Hub — Figure 1: Architectural data plane showing real-time streaming between @smoke-monkey/ui, the harness engine, and external MCP tools.

1. Frontend UI

@smoke-monkey/ui

Drop-in React chat interface: streaming markdown, collapsible tool cards, charts, and 14 theme presets.

2. Backend Engine

@smoke-monkey/harness

Autonomous execution loop with 6 phases, automated loop guards, 24 tools, and JIT skills discovery.

3. MCP Hub

@smoke-monkey/mcp

22 tools for scaffolding, planning, and verifying agents from Claude Code, Cursor, Codex, and Windsurf.

Minimal Autonomous Agent

Create an agent in 20 lines of code. Planning, tool invocation, and streaming are handled for you:

agent.tstypescript

1import { createAgent } from '@smoke-monkey/harness';
2
3// 1. Initialize the autonomous harness
4const agent = createAgent({
5  provider: 'nvidia', // Reference provider
6  model: 'nvidia/nemotron-3-super-120b-a12b',
7  workspacePath: process.cwd(),
8  autoApprove: true, // Execute safe tools automatically
9});
10
11// 2. Stream tokens in real time
12agent.on('text.delta', (e) => process.stdout.write(e.data.delta));
13agent.on('tool.started', (e) => console.log(`\n⚡ Tool invoked: ${e.data.toolName}`));
14
15// 3. Execute an autonomous engineering task
16const result = await agent.run('Inspect this codebase and count total lines of code in src/');
17console.log(`\n\n✅ Finished with status: ${result.status}`);