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.

Refactoring

9 min readUpdated: October 2026

Automated Codebase Refactoring with AI Agents: TypeScript Patterns & Safety

Technical Review: Smoke Monkey Core Architecture Team

|Tested on Node.js 18+ & Bun|TypeScript 5.x

Quick Answer & Executive Definition

Automated Codebase Refactoring with AI Agents: TypeScript Patterns & Safety: Designed as a zero-dependency, open-source TypeScript architecture under the MIT License with native Model Context Protocol (MCP) support and deterministic phase state machines.

Key Architectural Takeaways

Rename exports across hundreds of files with semantic understanding
Migrate from deprecated APIs to new versions automatically
Enforce consistent patterns across the entire codebase in one run
Run TypeScript compiler after each change to validate correctness

The Safe Refactoring Loop

Safe large-scale refactoring follows a validate-edit-validate cycle:

Baseline — Run tsc --noEmit and record the initial error count (should be 0).
Refactor — Agent edits files per the refactoring plan.
Validate — Run tsc --noEmit again and verify error count hasn't increased.
Test — Run the test suite and confirm all tests still pass.
Repeat — Continue with the next batch of files.

This loop ensures the agent never leaves the codebase in a broken state.

refactor-agent.tstypescript

1import { createAgent } from 'smoke-monkey-harness';
2
3const agent = createAgent({
4  provider: 'anthropic',
5  model: 'claude-3-7-sonnet',
6  workspacePath: process.cwd(),
7  autoApprove: true,
8});
9
10await agent.run(`
11  We are migrating from our old fetch wrapper (src/utils/api.ts → fetchData())
12  to the new typed client (src/lib/client.ts → apiClient.get/post/put/delete()).
13
14  Steps:
15  1. Find all usages of fetchData() across the codebase using grep.
16  2. Replace each usage with the equivalent apiClient method.
17  3. After every 10 files, run npx tsc --noEmit and fix any errors before continuing.
18  4. Run npm test at the end and fix any failing tests.
19  5. Report a summary of all changes made.
20`);

Common Migration Patterns

Dependency upgrades: Instruct the agent to read the migration guide for a major library version, find all usages of deprecated APIs, and update them to the new equivalents.

Naming convention enforcement: Ask the agent to rename all React components from PascalCase files to kebab-case filenames, updating all import paths accordingly.

Pattern extraction: Identify repeated inline logic (e.g., error handling boilerplate) and extract it into shared utilities, then replace all instances with calls to the new utility.

Google Search Questions & Answers

Frequently Asked Questions

Q:How do I handle refactoring that spans 500+ files?

Use the multi-agent orchestration pattern: a supervisor agent creates a file-by-file refactoring plan, then parallel worker agents process batches of 20-50 files concurrently.

Q:Can the agent handle refactoring that changes public API signatures?

Yes, but provide a clear migration contract: "function X(a: string) becomes X(options: { name: string }). Update all 47 call sites." The more explicit the contract, the more accurate the agent.

Related Alternatives & Comparisons

vs Cursor Composer

Cursor Composer Agent Alternative: Embeddable Code Editing Engine

Compare

vs Cognition Devin

Open Source Devin Alternative: Build Autonomous Software Engineers in TypeScript

Compare

Build with Smoke Monkey Harness

Zero dependencies. 24 built-in tools. Human-in-the-loop safety. 100% open source under the MIT License.

npm install smoke-monkey-harness