License & Deployment Mix: 14 tools – 11 OSS, 0 SaaS. (OSS and SaaS counts can overlap when an open-source tool also offers a vendor-hosted edition.)

Note: CLI/IDE clients of cloud LLM APIs.

What Are AI Development Tools?

AI development tools (also called AI coding assistants, AI pair programmers, or agentic coding tools) are software that uses large language models (LLMs) to help developers write, edit, debug, review, and ship code. They range from simple autocomplete engines to fully autonomous coding agents that can read a codebase, plan changes, write code across multiple files, run tests, fix failures, and create pull requests – all with minimal human intervention.

This category also includes orchestration and session management tools – software that sits above the coding agents, coordinating workflows, tracking sessions, managing checkpoints, and adding structured processes to otherwise free-form AI interactions.

AI development tools solve several problems for individual developers and engineering teams:

• Velocity – AI can generate boilerplate, implement well-understood patterns, and handle tedious refactoring much faster than a human
• Context switching – agents with large context windows can hold an entire codebase in memory and make consistent cross-file changes
• Knowledge augmentation – models trained on billions of lines of code can suggest idiomatic patterns for unfamiliar languages and frameworks
• Review and quality – AI can review PRs, flag security issues, suggest improvements, and run audits faster than manual review
• Onboarding – new team members can use AI assistants to explore and understand unfamiliar codebases

The information on these pages was researched by a combination of human review and large language models. To suggest an addition or correction, please contact us. Prepared by Rhodium Systems Inc., author of the ResorsIT platform — a unified IT operations management platform for IT teams and MSPs that integrates a curated suite of open-source, commercial, and SaaS applications into a single system with shared identity, single sign-on, access control, and a common audit trail. Use this catalogue only as a starting point for your own research, and review any tool carefully against your own requirements before relying on it. Catalogue data version 2026.197.

Comparison

These tools split into two interface classes. Terminal (CLI/TUI) agents – Claude Code, Gemini CLI, Aider, OpenCode, Crush – suit automated, scriptable, headless, and over-SSH work, where they integrate far better than a GUI. The non-terminal (GUI / IDE / web) tools – Cursor, Windsurf, GitHub Copilot, Continue, Cody, Kandev – favour interactive editing. The Interface table below records each tool’s class alongside its vi-mode and multi-repo support.

Tool Overview

Pricing Comparison

Capabilities Comparison

Extensibility & Integration

Deployment & Platform

Interface Class, Vi Mode & Multi-Repo Access

Each tool’s interface class (Terminal = CLI/TUI; Non-terminal = GUI / IDE / web), whether it ships a built-in vim / vi modal-editing mode, and whether it has controlled access to more than one repository at once. “No” vi mode = none shipped today (some have open requests). Multi-repo: Yes = multiple repos/dirs at once; controlled = with documented per-repo access control; Single = one repo at a time.

¹ gstack runs as slash-command skills inside a terminal host agent (Claude Code / Codex), so its interface and multi-repo reach are the host’s. ² Also ships a CLI (Cursor cursor-agent, Continue cn, Cody CLI, Copilot CLI), but the primary, access-controlled surface is the GUI / editor.

Interface class: the Terminal group (Claude Code, Gemini CLI, Aider, OpenCode, Crush; plus the gws / gstack / Entire utilities) suits headless and scriptable use; the Non-terminal group targets interactive editing.

Vi mode today: only Claude Code, Gemini CLI, Aider, and Cursor ship a built-in vi/vim modal-editing mode. Crush and OpenCode have open feature requests; Windsurf offers vim only via the VS Code VSCodeVim extension in its editor (not in the AI chat); Cody, Continue, and the Copilot CLI have none.

Multi-repo access: among the internal terminal agents, only Claude Code (--add-dir) and Gemini CLI (--include-directories) span several repos in one session – via a local per-directory allow-list, not org-scoped RBAC. Aider, OpenCode, and Crush are single-repo. Enterprise-grade controlled multi-repo access (indexing that respects code-host repo permissions) is a non-terminal-tool trait: Cursor, Windsurf, GitHub Copilot, and Cody.

SSO / Enterprise Features

Open Source Comparison

Best for data sovereignty: Aider + local model (Ollama), Crush + local model, Continue + local model. These combinations keep all code and prompts on-premises with no external API calls.

Cost & Usage Tracking

AI development tools are developer-workstation software, not server-side services, so there is no metrics endpoint to scrape. Cost and usage tracking still matters:

• API token costs – track spending across Anthropic, OpenAI, Google APIs via billing dashboards or API usage endpoints
• Entire – captures token usage per session; data stored in Git (can be extracted for reporting)
• Enterprise dashboards – Copilot, Cursor, and Claude Team/Enterprise provide admin usage dashboards
• Custom telemetry – gstack supports opt-in Supabase telemetry for sprint metrics

Tools

14 tools.

Aider

Aider is an open-source command-line AI pair programming tool. It runs in the terminal and lets developers pair program with any LLM to edit code in existing Git repositories.

License: Apache-2.0 (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Claude Code

Claude Code is Anthropic’s agentic coding assistant. It runs in the terminal (or IDE) and can autonomously read files, write code, execute shell commands, run tests, fix failures, create commits, and open pull requests.

License: MIT (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Cody (Sourcegraph)

Cody is Sourcegraph’s AI coding assistant. Its key differentiator is deep code search context – Cody leverages Sourcegraph’s code search and intelligence platform to retrieve context from across entire organizations’ codebases, not jus…

License: MIT (OSS) · Kind: library · Deploy: package · SSO: none

Website · Source

Continue

Continue is an open-source AI code assistant that has evolved from a pure IDE extension into a platform focused on “AI checks” – source-controlled agents that run as GitHub status checks on every pull request.

License: Apache-2.0 (OSS) · Kind: library · Deploy: package · SSO: none

Website · Source

Crush (formerly OpenCode)

Crush is a terminal-based AI coding assistant with a rich TUI (text user interface) built on Bubble Tea, the Go TUI framework from Charmbracelet.

License: MIT (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Cursor

Cursor is an AI-powered code editor built as a fork of VS Code by Anysphere, Inc. It embeds AI deeply into the editing experience with agents, Tab autocomplete, inline editing, and code review.

License: Proprietary (proprietary) · Kind: desktop · Deploy: native · SSO: none

Website

Entire

Entire is an AI coding session tracker and checkpoint manager that hooks into the Git workflow to capture everything AI agents do during development sessions – prompts, responses, files touched, token usage, and tool calls.

License: MIT (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Gemini CLI

Gemini CLI is Google’s open-source AI coding agent for the terminal. It brings Gemini models (Gemini 3 Pro and Flash) directly into the command line, providing intelligent coding assistance, code generation, codebase analysis, and automatio…

License: Apache-2.0 (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

GitHub Copilot

GitHub Copilot is a widely adopted AI coding assistant offering inline completions, chat, agent mode, and a cloud coding agent, powered by multiple models across popular IDEs.

License: Proprietary (proprietary) · Kind: library · Deploy: package, saas · SSO: none

Website

Google Workspace CLI (gws)

Google Workspace CLI (gws) is an open-source, unified command-line tool covering every Google Workspace API, with AI agent skills, an MCP server mode, and structured JSON output for automation.

License: Apache-2.0 (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

gstack

gstack is an orchestration framework that sits on top of Claude Code and transforms it into a “virtual engineering team” with specialized roles – CEO, engineer, designer, QA lead, security officer, and release manager.

License: MIT (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Kandev

Kandev is a self-hosted, server-first AI development environment that orchestrates 20+ coding agents in parallel behind a kanban board and IDE-like workspace.

License: AGPL-3.0 (OSS) · Kind: web · Deploy: docker, native · SSO: none

Website · Source

OpenCode

OpenCode is an open-source, provider-agnostic terminal AI coding agent (opencode.ai) with a client/server architecture, LSP and MCP support, built by Anomaly.

License: MIT (OSS) · Kind: cli · Deploy: native, package · SSO: none

Website · Source

Windsurf

Windsurf is an AI-powered IDE originally created by Codeium. OpenAI acquired the product line, and it is now operated by Cognition, Inc. It markets itself as “the most intuitive AI coding experience, built to keep you in flow.”

License: Proprietary (proprietary) · Kind: desktop · Deploy: native · SSO: none

Website