← Directory

Fletch

Open Source

• Updated Aug 1, 2026, 03:05 AM

Fletch is a native macOS integrated development environment tailored for building and managing AI coding agents.

Use Cases ▼

Safely exploring codebase modifications with AI assistance

Automated refactoring with mandatory approval gates

Generating and testing patches in isolated environments

Auditing AI-generated changes before merging

Learning codebase structure via shared symbol and call-graph index

Rapid prototyping with AI suggestions under user supervision

What Problem It Solves ▼

Fletch addresses the trust and safety challenges in AI-assisted software development by isolating agent execution in sandboxed environments, sharing deep code structure through a unified index, and enforcing testing or approval checkpoints to prevent unintended or harmful code changes.

Key Features ▼

✓

Strong sandboxing prevents agent misbehavior

✓

Integrated testing/approval gates enhance safety and code quality

✓

Native macOS performance and user experience

✓

Shared code index improves agent contextual understanding

✓

Open source with permissive licensing

Overview

Fletch is a native macOS integrated development environment tailored for building and managing AI coding agents. It enforces strong isolation by sealing each agent in its own repository clone, sandboxed with macOS Seatbelt or Docker, preventing unauthorized access. Agents access a shared symbol and call-graph index served over the Model Context Protocol (MCP), providing a comprehensive understanding of the codebase. Every action an agent takes is gated: it must pass automated tests or obtain explicit user approval before proceeding, ensuring safety and reliability. The IDE integrates with Anthropic's Claude model and is designed for developers who want to leverage AI assistance while maintaining strict control over code changes and execution environment.

Pros & Cons

Pros

Strong sandboxing prevents agent misbehavior

Integrated testing/approval gates enhance safety and code quality

Native macOS performance and user experience

Shared code index improves agent contextual understanding

Open source with permissive licensing

Cons

−

macOS only, limiting cross-platform adoption

−

Requires manual approval steps, which can slow automation

−

Currently limited to Claude model integration

−

Setup and configuration may have a learning curve

−

Dependency on Docker for containerized isolation

Similar Tools

Cursor Windsurf GitHub Copilot Workspace Devin Aider Tabby

More in AI Agents

ECC

Open Source ★278,144

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

Hermes Agent

Open Source ★253,278

The agent that grows with you

Firecrawl

Open Source ★190,975

Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥

Everything Claude Code

Open Source ★182,056

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

Rea

Open Source ★179,539

Reverse engineer anything with agents, from app behavior down to native binaries.

Ponytail

Open Source ★162,150

Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.