plandex

Open source AI coding agent. Designed for large projects and real world tasks.

open-sourcecoding-agents
15.2k
Stars
+120
Stars/month
0
Releases (6m)

Star Growth

+26 (0.2%)
14.9k15.2k15.5kMar 27Apr 1

Overview

Plandex is an open-source, terminal-based AI coding agent specifically designed for tackling large, complex development tasks that span multiple files and require numerous steps. Unlike typical AI assistants that work with single files or simple requests, Plandex excels at planning and executing comprehensive coding projects that might involve refactoring dozens of files, implementing multi-component features, or modernizing entire codebases. The tool can handle up to 2 million tokens of context directly (approximately 100,000 tokens per file) and can index much larger projects containing 20 million tokens or more using advanced tree-sitter project mapping. This massive context window allows it to maintain awareness of complex project structures and dependencies throughout the development process. With over 15,000 GitHub stars, Plandex has gained significant traction among developers working on substantial codebases who need an AI assistant capable of understanding and modifying large-scale projects systematically. The tool operates through a cumulative diff approach, building changes incrementally while maintaining project coherence.

Deep Analysis

Capabilities

  • Terminal-based AI coding tool for planning and executing large multi-file coding tasks
  • 2M token effective context window with smart file loading; supports 20M+ token projects via tree-sitter maps
  • Sandboxed cumulative diff review keeping AI changes separate until explicit approval
  • Multi-model support combining Anthropic, OpenAI, Google, and open-source providers
  • Configurable autonomy from full automation to granular manual control
  • Automated terminal command debugging for builds, tests, and linters
  • Browser application debugging with Chrome integration

🔗 Integrations

Anthropic ClaudeOpenAIGoogle modelsOpenRouter.aiTree-sitter (30+ languages)DockerGit

Best For

  • Developers working on large-scale multi-file refactoring or feature implementation
  • Terminal-centric developers who prefer CLI over IDE plugins
  • Teams needing sandboxed AI code changes with explicit review before applying

Languages

Go

Deployment

CLI installationSelf-hosted server modeBring-your-own API keys

Pricing Detail

Free: Open-source, self-hosted with own API keys
Paid: Plandex Cloud discontinued (10/2025) — self-host only

Known Limitations

  • Terminal-only — no GUI or IDE plugin
  • Plandex Cloud shutting down — requires self-hosting
  • Learning curve for REPL-based workflow
  • Requires API keys from model providers

Pros

  • + Exceptional context handling with 2M+ token capacity for understanding large, complex codebases
  • + Purpose-built for real-world, multi-file projects rather than simple single-file tasks
  • + Open-source with self-hosting options, providing full control over your development environment

Cons

  • - Terminal-based interface may not appeal to developers who prefer GUI tools
  • - Potentially overkill for simple, single-file coding tasks or quick fixes
  • - Requires setup and configuration that may be complex for casual users

Use Cases

  • Large-scale refactoring projects that touch dozens of files across a codebase
  • Implementing comprehensive features that require changes across multiple components and layers
  • Modernizing legacy codebases with systematic updates and architectural improvements

Getting Started

1. Install Plandex using the 30-second install process from the official website or docs. 2. Configure your environment and connect to either the hosted service or set up local self-hosted mode. 3. Initialize your first project by pointing Plandex at your codebase and describing a large task you want to accomplish.

Compare plandex