An open atlas of developer tools
Explore developer tools, compare the details, and learn from real experience.
Catalog
1,300 developer tools across AI, design, development, and more. Start with the latest additions.
catalog / recommended
Editor-selected tools from the public catalog. Review scores and evidence status are disclosed separately, so a recommendation does not imply a Verified hands-on review unless the linked record says so.
Agentic coding tool by Sourcegraph (formerly Cody)
Amp is a multi-model coding agent for terminal, web, macOS/iOS, and IDE-connected workflows. Vendor docs describe the same agent and threads across surfaces, with remote Orbs, local Runners, Dial modes (low / medium / high / ultra), shared threads, MCP, plugins, and BYOK / linked ChatGPT subscriptions. Pricing spans a free Hobby tier, optional Megawatt ($20/mo) and Gigawatt ($200/mo) Individual plans, no-extra-charge Teams workspaces, pay-as-you-go credits, and Enterprise.
Anthropic's agentic coding CLI
Anthropic's agentic CLI coding tool that delegates complex tasks to Claude directly from the terminal. Understands entire codebases via automatic context gathering, edits multiple files, runs shell commands, and manages Git workflows autonomously. Supports CLAUDE.md for persistent project instructions, integrates with VS Code and JetBrains, and uses Claude Opus/Sonnet with extended thinking for complex architectural decisions. Built for terminal-first developers.
xAI's terminal coding agent with parallel subagents and worktree-aware automation
Grok Build is xAI’s terminal-first coding agent for planning, editing, testing, and reviewing code from a local CLI. It exposes subagent controls, worktree mode, headless JSON output, best-of-N parallel attempts, sandbox profiles, and cross-session Memory (/memory, /dream) for durable project conventions and facts. It fits developers comparing Claude Code, Codex, and Gemini CLI for local agentic workflows with deeper parallel execution.
Persistent AI teammates for cross-app work
Grok Bot is SpaceXAI's beta workspace for persistent AI teammates that use apps, websites, files, terminals, plugins, and MCP on a user-scoped cloud computer. Bots can learn routines, coordinate in parallel, and, when enabled by team policy, launch Cursor Cloud Agents that work on code in isolated environments and return pull requests and verification artifacts.
The AI-first code editor
AI-first code editor built as a VS Code fork that deeply integrates LLMs into every part of the development workflow. Features Tab autocomplete with multi-line predictions, Cmd+K inline editing, AI chat with full codebase awareness, and Agent mode for autonomous multi-file edits with terminal execution. Supports GPT-4, Claude, and more with automatic context from project files and docs. Includes privacy mode for SOC 2 compliance. The leading AI-native IDE with 100K+ paying users.
OpenAI coding agent for app, editor, terminal, and cloud work
Codex is OpenAI's coding agent for software development across the Codex app, editor, terminal, and cloud tasks. It helps write, review, debug, refactor, and automate code, with ChatGPT plan access for managed surfaces and API-key usage for CLI, SDK, and IDE workflows. The open-source CLI and SDK support local repository work, while cloud features add GitHub review, Slack/Linear integrations, worktrees, skills, MCP, and automations.
editorial / latest
Long-form reviews, head-to-head comparisons, and completed weekly diffs with visible sources and verification dates. Reviews disclose when hands-on evidence is complete, partial, missing, or still unclassified.
Goose vs OpenCode: Local MCP Agent or Open-Source Coding Agent?
Goose and OpenCode both run open-source AI agents that help developers ship work from the terminal, but they emphasize different product shapes. Goose is an open-source, model-agnostic local agent from the Agentic AI Foundation (AAIF) lineage — Desktop, CLI, and API — built around MCP extensions, recipes, and BYOK or local Ollama. OpenCode is an open-source coding agent spanning a terminal interface, IDE extension, and desktop app, with any-provider models and a polished coding-agent loop. Use Goose when you want a privacy-forward local agent with deep MCP extensibility across code and broader workflows. Use OpenCode when you want a coding-agent surface across terminal, IDE, and desktop with strong day-to-day developer experience. Existing aicoolies Scores: Goose overall 84, speed 80, privacy 95, developer experience 78 (tested 2026-03-29); OpenCode overall 83, speed 81, privacy 90, developer experience 86 (tested 2026-09-05). Overall is close; choose on product shape first, and treat the scoreboard as secondary review evidence with those dates.
Claude Code vs Trae Agent — Anthropic Terminal Agent or ByteDance Open-Source BYOK Agent?
Claude Code and Trae Agent both run coding agents from the terminal, but they are different product shapes. Claude Code is Anthropic’s commercial terminal-native agent — deep Claude reasoning, multi-file edits, shell and git loops, multi-agent coordination, and CLAUDE.md project memory, with Pro / Max / Team or API billing. Trae Agent is ByteDance’s MIT open-source software-engineering agent built to swap LLM providers (OpenAI, Anthropic, Doubao, Gemini, Azure, Ollama) on a BYOK path. Use Claude Code when you want Anthropic’s recommended terminal-agent path and a polished commercial CLI. Use Trae Agent when you want an inspectable, provider-agnostic agent you can run against your own gateway or local Ollama. Existing aicoolies Scores: Claude Code overall 92, speed 88, privacy 65, developer experience 94 (tested 2026-03-27); Trae Agent overall 81, speed 78, privacy 86, developer experience 79 (tested 2026-05-04). Choose on commercial Anthropic CLI vs open provider-agnostic agent fit; treat the scoreboard as secondary review evidence with those dates, not the whole decision. Do not confuse Trae Agent with ByteDance’s separate AI-native IDE Trae (/tools/trae). This page is terminal SWE-agent vs terminal SWE-agent — not an IDE bake-off, and not Claude Code vs Jules.
Google Antigravity vs Zed — Google Agent-First Platform or Rust Native Editor with AI?
Google Antigravity and Zed both sit in the IDE / code-editor stack, but they solve different jobs. Antigravity is Google’s agent-first development platform — desktop/IDE plus the Antigravity CLI (agy) — built around planning and implementation agents with Gemini-class and multi-model options. Zed is an open-source, Rust-built, GPU-accelerated editor focused on native speed and real-time collaboration, with Agent Panel, Edit Prediction, Inline Assistant, and ACP paths for external agents plus BYOK. Existing Scores: Antigravity overall 88 (tested 2026-09-05); Zed overall 85 (tested 2026-03-27). Choose on agent-platform vs performance-editor fit; treat the scoreboard as review evidence, not the whole decision.
Grok Build vs Grok CLI — Official xAI Agent or Community Grok Terminal?
Grok Build and Grok CLI both put Grok models in a terminal workflow, but they are not the same product. Grok Build is xAI’s official terminal-first coding agent — SuperGrok / X Premium+ interactive access or metered xAI API headless — with subagents, worktrees, and Memory (vendor 16 September 2026: durable markdown notes with /memory and /dream). Grok CLI is a community MIT-licensed command-line agent for Grok API keys: lightweight, inspectable, and scriptable without xAI’s subscriber-linked Build surface. Use Grok Build when you want official xAI support and the Build agent desk. Use Grok CLI when you want an open-source Grok API terminal you can fork and script. Existing Score references both land at overall 82 (Build Score tested 2026-05-28; CLI Score tested 2026-07-02). Choose on official vs community fit.
rooms
Context
I’ve been using Amp’s Orbs to send development tasks to the cloud, and the value of the $20/month Megawatt tier has been impressive. An Orb is a remote, isolated machine for an…
Which option gives you the best balance of reliability, parallelism, and predictable cost?
Context
I’ve been using Devin in the cloud for quite a while, but Fusion in Devin Desktop and CLI has changed how I use it locally. I can now pair Fable 5.1 or GPT-6 Astra as the lead…
How has it performed on larger real-world tasks, and how do its reliability and quota consumption compare with running the frontier model alone?
community / stack builders
Developers building their AI toolchains in public.
field notes / latest
Using Devin Max across desktop and web, I delegated a cloud task and merged a PR. Screenshots, recordings, Fusion routing, and integrated review made the handoff easier to follow; ACP support offers a shared interface for other agents.
Read the Field Note →catalog / how it works
The shared rubric and evidence states behind scores, source checks, and verification dates.
scoring
aicoolies Score v1 defines overall quality, speed, privacy, and developer experience on a public 0–100 rubric. Reviews marked Verified also expose the test date, version, environment, artifact, and source-check dates; older scores are labeled separately.
freshness
Pricing and plan limits move constantly. A page shows a source-check or last-verified date only when that editorial check was recorded; routine CMS updates are not presented as verification.
honesty
Discontinued products leave the recommendations and move to the Graveyard with the date they shut down and what replaced them, instead of quietly rotting in a list.
open data
The catalog is published as CC-BY 4.0 JSON, an OpenAPI-described API, and an MCP server. Supported public catalog and guide URLs also answer in markdown on request, so agents can read the same published facts without scraping presentation markup.
build with us
Listings are free. The dataset is CC-BY 4.0. Agents can read the same published facts without scraping the page.
aicoolies Score v1 defines four 0–100 axes: overall, speed, privacy, and developer experience. Each review now discloses whether its test evidence is Verified, Partial, Not yet evidenced, or Legacy, and the full rubric is published on the methodology page.
Records display source-check and last-verified dates only when those checks have been completed. Missing verification metadata is left visibly unclassified rather than inferred from a routine edit. Confirmed discontinued tools can move to the Graveyard instead of lingering as live recommendations.
Yes. The tool, comparison, and review datasets are published as JSON under CC-BY 4.0, so you can use them with attribution to aicoolies.com. There is no API key and no signup.
Yes. Supported public catalog and guide URLs answer in markdown when requested, and the catalog is published as an OpenAPI-described JSON API and an MCP server. Public reads need no token at all.
No. Listings, scores, and recommendations cannot be bought and outbound tool links carry no affiliate parameters. aicoolies does sell clearly labeled sponsored placements (a sponsor rail and one sponsor per category); those never affect scores, verdicts, or rankings. The scoring rubric is published precisely so a ranking can be argued with instead of taken on trust.
A single editor, working from Izmir, Turkey, publishes the catalog under a named byline. The editorial team page lists who assigns the scores and how to reach them, and corrections sent with a source land in the next editorial pass.
Submit it with links to the official documentation and pricing. An editor checks the supplied claims and writes the entry; a review is called hands-on or Verified only when its reproducible test evidence is complete. Listings are free either way: aicoolies does not sell rankings, editorial recommendations, or removal from the Graveyard; sponsored placements are labeled and separate.
