# DevThrottle

> DevThrottle is a control room for running and monitoring AI coding agents. This file points LLM crawlers to the highest-signal pages and machine-readable data.

Full long-form product reference (definition, components, pricing, privacy, FAQ): https://devthrottle.com/llms-full.txt

## Core pages

- [DevThrottle](https://devthrottle.com/): Windows app that runs and supervises multiple command-line coding agents side by side and shows you who needs you. Open source (MIT). Download it at https://devthrottle.com/download - no account needed to install; you sign in when you connect a gateway.
- [Install with an AI coding agent](https://devthrottle.com/docs/install#install-agent): If an AI coding agent (Claude Code, Codex) is doing the install, it should NOT download and run the graphical installer. It downloads the headless CLI installer devthrottle-setup-cli-win-x64.exe from the latest GitHub release (https://github.com/thefrederiksen/devthrottle/releases/latest), verifies its SHA-256 against release-manifest.json, and runs "devthrottle-setup-cli-win-x64.exe install" (per-user, no admin), then signs in with "devthrottle-setup-cli-win-x64.exe signin" (or "devthrottle-setup-cli-win-x64.exe enroll" to join a gateway you already run). Same install engine as the wizard; the only human step is that browser sign-in.
- [Pricing](https://devthrottle.com/pricing): Pro is $29/mo founding price ($49 list): a hosted gateway, dictation, spoken replies, and the wingman - a brain that watches your agents. Pro self-host is $24/mo founding ($39 list): the same AI on your own gateway. Max (the fleet brain) is announced, not yet for sale. Enterprise is per seat. Every new account starts with 14 days of Pro, no credit card. Dictation, spoken replies, and the wingman are included in Pro and the trial with no usage charge. The app itself is open source (MIT) and costs nothing to self-host. Prepaid credits are separate and apply only to direct catalog-model API calls.
- [Open source](https://devthrottle.com/open-source): The full app is MIT licensed - source, releases, and the do-it-yourself gateway path on GitHub (https://github.com/thefrederiksen/devthrottle).
- [AI Coding Agents Directory](https://devthrottle.com/coding-agents): Filterable source-backed catalog of 45 AI coding agents.
- [AI Coding Agents Matrix](https://devthrottle.com/coding-agents/matrix): Side-by-side comparison of platform, pricing, autonomy, source status, and product status.
- [AI Coding Agents Leaderboard](https://devthrottle.com/coding-agents/leaderboard): Public adoption-signal ranking with source confidence.
- [Agent Knowledge Base](https://devthrottle.com/docs/agent-knowledge-base): How DevThrottle researches, verifies, and maintains the catalog.
- [Machine-readable catalog](https://devthrottle.com/coding-agents.json): JSON feed for all cataloged agents and source links.

## Documentation

Every documentation page is also served as plain markdown at its URL with ".md" appended (e.g. https://devthrottle.com/docs/install.md). The full text of all docs pages is concatenated into https://devthrottle.com/llms-full.txt.

- [Docs home](https://devthrottle.com/docs): All DevThrottle documentation - install, guides, and reference.
- [Get started with DevThrottle](https://devthrottle.com/docs/start): The numbered path through DevThrottle - read the steps in order, one sitting each.
- [Step 1: Install DevThrottle and connect a gateway](https://devthrottle.com/docs/start/install): Create your account, install, connect a gateway, and get out. Using the Director is step 2.
- [Step 2: Start your first session](https://devthrottle.com/docs/start/first-session): From an empty Director to one agent doing real work in your own code: the New Session window end to end, then the first instruction and the first minute of watching.
- [Step 3: The prompt bar is not the terminal](https://devthrottle.com/docs/start/prompt-bar): The agent has its own live input line; DevThrottle gives you a second box outside it that holds what you write until you send it.
- [Step 4: Steer a session that is going wrong](https://devthrottle.com/docs/start/steering): Take the lowest rung that works: stop the turn, correct it while the conversation is intact, and keep Compact and Clear context for when the conversation itself is the problem.
- [Step 5: Work with pictures](https://devthrottle.com/docs/start/pictures): An agent reads a picture as a file on the machine it runs on, so every route ends the same way: the file on that machine, and its path in the prompt.
- [Step 6: See what changed](https://devthrottle.com/docs/start/what-changed): An agent tells you what it did; the repository holds what happened. Read the second before believing the first, then keep it or send it back.
- [Step 7: Run several sessions at once](https://devthrottle.com/docs/start/several-at-once): Three sessions at once is a different job from one: you read a list of coloured squares, answer the longest wait, and queue for whichever one is busy.
- [Step 8: Where else your fleet lives](https://devthrottle.com/docs/start/fleet-elsewhere): Your agents are not trapped in the window you started them in: the same sessions, the same colours, in a browser on any machine and on the phone in your pocket.
- [Install DevThrottle in three steps](https://devthrottle.com/docs/start/easy-install): Account, install, first agent. Three steps, on Windows or Mac.
- [What DevThrottle is](https://devthrottle.com/docs/overview): The one idea: every coding agent in one place, and it tells you who needs you.
- [Install DevThrottle and run your first agent](https://devthrottle.com/docs/install): The install routes in full - Windows, the Mac Terminal command, and the headless installer an AI coding agent runs.
- [DevThrottle tutorials](https://devthrottle.com/docs/tutorials): Every DevThrottle tutorial, grouped by the job you are trying to do and by where you are working.
- [Talk to your agents instead of typing](https://devthrottle.com/docs/tutorials/dictation): Nothing appears while you talk, and that is deliberate: press Pause to read it back, then Insert to check it or Send to let it go.
- [Working with pictures](https://devthrottle.com/docs/tutorials/pictures): Three nested sets of picture types - five in the desktop Screenshots panel, those five plus webp in the Expand preview and the browser gallery, those six plus heic on the way in - so one folder looks different depending on where you stand. Plus the panel buttons, the folder setting, and the boundary.
- [Roles: who plans, who manages, who builds](https://devthrottle.com/docs/tutorials/roles): Declare an Architect, hand work down and watch the Worker badge appear on its own - the declared planner keeps its A - then read the colours: a supervised Worker never turns red; its Supervisor deals with it.
- [One agent starting another](https://devthrottle.com/docs/tutorials/agent-starts-agent): From inside a running session, spawn hands work to a fresh session - a Worker supervised by its spawner that never turns red at you - message ask collects the answer, and session done reaps the helper.
- [Repositories and worktrees](https://devthrottle.com/docs/tutorials/repositories-and-worktrees): Two sessions in one folder edit over each other and nothing stops them. Give each its own worktree, start the session there, and see every worktree of a repository from the Worktrees page.
- [Workflows: your named ways of working](https://devthrottle.com/docs/tutorials/workflows): Read what your fleet already works by, pull one down, write the rules you keep repeating, and publish it everywhere at once.
- [Line up the next instruction](https://devthrottle.com/docs/tutorials/queue-next): Write the next instruction while the agent works, queue it with Ctrl+Shift+Enter, order the ones you have stacked up, and send one yourself when the agent stops - the queue holds, it does not deliver.
- [Pair your phone with your fleet](https://devthrottle.com/docs/tutorials/pair-phone): Scan the code on the Cockpit's Phone page, approve the phone on devthrottle.com, and turn on notifications - five minutes at your desk, done before the day you need it.
- [Answer an agent in thirty seconds](https://devthrottle.com/docs/tutorials/quick-answer): The roster puts what is waiting on you at the top in red. Read the question, answer short, move on.
- [A whole session by voice, on a walk](https://devthrottle.com/docs/tutorials/voice-walk): Put the fleet on voice, tick Auto-speak, and the queue reads itself to you - Respond by talking, Snooze the rest, and the three-second pause is the way out.
- [Your whole fleet in a browser tab](https://devthrottle.com/docs/tutorials/cockpit-anywhere): Reach every session from any browser, put what is waiting on top, answer it, and sign out again.
- [The fleet map](https://devthrottle.com/docs/tutorials/fleet-map): Read the whole fleet in one picture, regroup it around the question you actually have, and click straight through to the session that is waiting.
- [What did I actually get done this week?](https://devthrottle.com/docs/tutorials/weekly-review): Widen the window to a week, read the work by repository, and turn the needed-you count into one prompt worth rewriting.
- [Teach it the words you use](https://devthrottle.com/docs/tutorials/dictionary): Add your vocabulary and the wrong spellings you actually saw, save, and the cleanup pass corrects them in the transcriptions it covers.
- [Your Throttle: what you have used](https://devthrottle.com/docs/tutorials/your-throttle): The private dashboard of your own driving: the three rings, your working day hour by hour, and the ranked repos and agents behind it.
- [Give an agent its own browser](https://devthrottle.com/docs/tutorials/agent-browser): A profile of its own: create it, sign it in once by hand, start it, and an agent on that machine drives it through Browser Harness - holding exactly the accounts you gave it.
- [Work that runs without you](https://devthrottle.com/docs/tutorials/scheduled-work): Create a recurring job, read the schedule back in plain English, and prove it works before two in the morning does.
- [Using more than one machine](https://devthrottle.com/docs/tutorials/many-machines): The Gateway is the only shared part. Point a second machine at it and that machine becomes somewhere you send work.
- [Driving DevThrottle from a script](https://devthrottle.com/docs/tutorials/scripting): Read the fleet as text, message a session, let a program ask through the Gateway, then hand it to the scheduler.
- [The Director](https://devthrottle.com/docs/director/overview): A tour of the Director window - the session sidebar, terminal pane, prompt bar, and toolbar.
- [Session cards and states](https://devthrottle.com/docs/director/sessions): How to read a session card: the status square colors, activity labels, timers, and badges.
- [Session roles and missions](https://devthrottle.com/docs/director/roles): Standalone, Architect, Manager, Worker, and the not-yet-recordable Reviewer - most worked out for you, any of the four declarable - plus missions the related sessions attach to.
- [Starting and steering agents](https://devthrottle.com/docs/director/steering): Start sessions, answer agents, queue prompts, and use Stop, Interrupt, Compact, and Clear context.
- [The embedded terminal](https://devthrottle.com/docs/director/terminal): The live terminal, the prompt bar (Send / Queue / Speak, slash commands), selection and paste, Reset View and Capture.
- [Source control and history](https://devthrottle.com/docs/director/source-control): The changes badge, the Source Control tab, and conversation history.
- [Repositories and GitHub](https://devthrottle.com/docs/director/repositories): Root folders, the repository list and how it is ordered, and what one repository's screen shows.
- [Agents and models](https://devthrottle.com/docs/director/agents): The Settings > Agents editor (detect, quick check, guided vs custom launch, model picker) and the memory/output-style slash commands.
- [Workspaces](https://devthrottle.com/docs/director/workspaces): Save Workspace, Load Workspace (replaces the current sessions), and what a workspace stores.
- [Renaming and resuming sessions](https://devthrottle.com/docs/director/session-lifecycle): Rename, /resume and /continue, and clean close behavior.
- [Director settings](https://devthrottle.com/docs/director/settings): The Director Settings dialog tab by tab, plus where to find the settings it does not hold.
- [What the Gateway does](https://devthrottle.com/docs/gateway/overview): The always-on hub your Directors dial out to - what lives on it, how your other devices connect, and where the gateway can run.
- [Keeping agents running](https://devthrottle.com/docs/gateway/lifecycle): What keeps running when - viewers vs the Director vs the Gateway, and restarts.
- [Connecting machines without opening a port](https://devthrottle.com/docs/gateway/portless): Every machine dials out and nothing listens - the standard connection since 2.0, with no port to open or forward.
- [Why the app feels slow at home (and how to make it fast)](https://devthrottle.com/docs/gateway/network): Relay warm-up vs direct path, the Diagnostics page, agent-run diagnosis, and the fix checklist.
- [Scheduled runs with the Gateway cron](https://devthrottle.com/docs/gateway/cron): Schedule unattended agent runs with the built-in cron - the Schedule page and the REST API.
- [Workflows: the rules your fleet works by](https://devthrottle.com/docs/gateway/workflows): One catalog of named ways of working, shared fleet-wide - authored as a draft, published everywhere at once, with running jobs pinned to their version.
- [Skills: the capabilities your agents reach for](https://devthrottle.com/docs/gateway/skills): One central library of capabilities, published fleet-wide in a moment - up to forty named in a session's briefing, the full instructions fetched only on use.
- [Running agents on more than one machine](https://devthrottle.com/docs/gateway/machines): Adding a machine to the fleet, per-Director health badges, and starting sessions on it remotely.
- [Sessions talking to sessions](https://devthrottle.com/docs/gateway/fleet-messaging): message send / message ask / session spawn, and the safety rails: team scope, human grants, rate limits.
- [Control API](https://devthrottle.com/docs/gateway/control-api): The Gateway is the programmatic surface: cc-devthrottle, the session key, and what an agent may call. No local port.
- [The Cockpit](https://devthrottle.com/docs/cockpit/overview): Your fleet in a browser: full controls, every machine in one place, served by your own Gateway.
- [The session rail](https://devthrottle.com/docs/cockpit/sessions): The roster: session numbers, card states, My order vs Attention first, New session, and the session menu.
- [Working in a session](https://devthrottle.com/docs/cockpit/session-view): Terminal, Chat, Voice, and Source Control tabs; Stop, Interrupt, Compact, Clear context, History; Send, Speak, Queue, Attach; and the dock.
- [The Fleet Map](https://devthrottle.com/docs/cockpit/fleet-map): The fleet as a map: eight ways to group it, the title search, and the per-Director Wobbly, Offline, and Not running states.
- [The Assistant](https://devthrottle.com/docs/cockpit/assistant): Ask your fleet a question, by typing or by voice: what is open, what is stuck, which machines are on - and act on the answer.
- [Answering agents remotely](https://devthrottle.com/docs/cockpit/remote): The away-from-your-desk flow: open the Cockpit, sign in, answer the waiting card.
- [DevThrottle on mobile](https://devthrottle.com/docs/mobile/overview): A quiet dot on the app icon when a session needs you; answer by voice or thumb from wherever you are.
- [Pairing your phone](https://devthrottle.com/docs/mobile/pairing): The one-time approval flow, and why what the phone keeps is a per-device key, not your account.
- [Working with sessions from your phone](https://devthrottle.com/docs/mobile/sessions): The roster, starting sessions, Chat with Send/Speak/Attach, the live terminal, and the file viewer.
- [Voice Mode on your phone](https://devthrottle.com/docs/mobile/voice): Spoken narration of completed turns, the three ways to turn voice on, and every answer the screen can give when it is quiet.
- [Voice](https://devthrottle.com/docs/voice/overview): Dictation everywhere you steer agents, with a custom dictionary that reaches every one of those surfaces.
- [The Voice Recorder](https://devthrottle.com/docs/voice/recordings): Long-form recording on the phone, transcribed on your Gateway, read and kept in the Cockpit.
- [Your dictation dictionary](https://devthrottle.com/docs/voice/dictionary): The Cockpit Dictionary page: Vocabulary terms and Common mistranscriptions, and how the cleanup pass applies them.
- [AgentEyes: a screen and audio recorder for Windows](https://devthrottle.com/docs/agent-eyes/overview): What AgentEyes is, what runs on your machine, and what uses DevThrottle hosted AI.
- [Recording with AgentEyes](https://devthrottle.com/docs/agent-eyes/recording): Presets, the camera, capture shortcuts, the library and the walkthrough builder - what works today.
- [Import an external video](https://devthrottle.com/docs/agent-eyes/import): Turn an outside video such as a Teams recording into a transcribed library recording.
- [Multilingual subtitles](https://devthrottle.com/docs/agent-eyes/subtitles): Translate a transcript into another language and burn the captions into a new video.
- [The agenteyes CLI](https://devthrottle.com/docs/agent-eyes/cli): Every agenteyes command and its options, straight from the tool.
- [The AgentEyes Control API](https://devthrottle.com/docs/agent-eyes/control-api): The loopback REST API on 127.0.0.1:7882 - every endpoint, for local automation.
- [Your account dashboard](https://devthrottle.com/docs/account/overview): One glance at your devices and subscription, and the map of the account pages.
- [Devices](https://devthrottle.com/docs/account/devices): The device list, the approval rule (nothing registers silently), the address list, and revoke as the kill switch.
- [API keys](https://devthrottle.com/docs/account/api-keys): Keys for the things you build: shown once, one per integration, activity attributed per key.
- [Usage](https://devthrottle.com/docs/account/usage): Your API activity, visible: by service, by model, and by key, with per-request detail on a key's own view.
- [Billing](https://devthrottle.com/docs/account/billing): Your subscription and trial state; credits are separate and only for direct catalog-model API calls.
- [Settings](https://devthrottle.com/docs/account/settings): Who you are to DevThrottle and what it sends you: nickname, the three email groups, product usage data.
- [DevThrottle API](https://devthrottle.com/docs/api/overview): One dt_ key, OpenAI-compatible surface - what is available today and how it is paid.
- [Authentication](https://devthrottle.com/docs/api/authentication): Bearer dt_ keys: header shape, SDK setup, shown-once rule, one key per integration.
- [Chat completions](https://devthrottle.com/docs/api/chat): OpenAI-shaped chat with SSE streaming; unknown models are hard errors, failures are never debited.
- [Transcription](https://devthrottle.com/docs/api/transcription): Multipart audio in, transcript out - included with Pro and the 14-day Pro trial.
- [Speech](https://devthrottle.com/docs/api/speech): Text to speech on the same dt_ key - streamed audio, catalog voices, included rather than credit-billed.
- [Models](https://devthrottle.com/docs/api/models): The source of truth for what your key can call - fetch it, do not hard-code ids.
- [Errors and limits](https://devthrottle.com/docs/api/errors): OpenAI-shaped errors; design for 402 up front; failed requests are never debited.
- [The cc-* CLI tools](https://devthrottle.com/docs/cli/overview): Eight shipped tools - documents, email, images, vault, fleet - installed, health-checked and repairable by the app itself.
- [CLI tool reference](https://devthrottle.com/docs/cli/reference): All eight shipped tools - converters, fleet CLI, email, images, vault - usage and key commands.
- [Supported coding agents](https://devthrottle.com/docs/agents/supported): Eight built-in drivers plus a custom type for any CLI, the permission setting each one launches with, and what a custom session does not get.
- [Glossary](https://devthrottle.com/docs/reference/glossary): Every DevThrottle term, defined in a sentence and linked to its page.
- [Troubleshooting](https://devthrottle.com/docs/reference/troubleshooting): Real failure modes with real fixes; the status rail is the first diagnostic.
- [Run DevThrottle on a Mac unattended](https://devthrottle.com/docs/reference/mac-unattended): Never let the screen lock - the checklist that keeps a Mac launching agents unattended, and what that posture costs in security.
- [What DevThrottle reports home](https://devthrottle.com/docs/reference/telemetry): No telemetry program any more; your conversations are pushed to your Gateway and kept 90 days.
- [Changelog](https://devthrottle.com/docs/reference/changelog): Release highlights here; the GitHub releases feed is the authoritative changelog.

## Priority CLI coding agents

- [Aider](https://devthrottle.com/coding-agents/aider): An open-source AI pair programming CLI that works in any terminal, makes git-committed edits across multiple files, and supports over 100 LLM providers.
- [Antigravity CLI](https://devthrottle.com/coding-agents/antigravity-cli): Google's current terminal agent path for individual Gemini coding workflows after the June 18, 2026 Gemini CLI transition, with migration support from Gemini CLI extensions and agent workflows.
- [Amp](https://devthrottle.com/coding-agents/amp): Sourcegraph's agentic coding tool that combines a terminal CLI and VS Code extension with deep codebase search, giving the model precise context retrieval across large repositories.
- [Claude Code](https://devthrottle.com/coding-agents/claude-code): A terminal-based agentic coding tool that runs in the developer's own environment, giving Claude read/write access to the file system, shell, and browser to complete coding tasks end to end.
- [Codex CLI](https://devthrottle.com/coding-agents/codex-cli): OpenAI's open-source agentic coding CLI that runs in the developer's terminal with sandboxed code execution, giving GPT and o-series models access to files and shell commands.
- [Cursor CLI](https://devthrottle.com/coding-agents/cursor-cli): Cursor's terminal agent surface for interactive and headless coding workflows, scripts and automation, MCP integration, shell mode, and cloud handoff from the command line.
- [GitHub Copilot CLI](https://devthrottle.com/coding-agents/github-copilot-cli): GitHub's terminal-native Copilot coding agent for local sessions, custom agents, sandboxed workflows, and GitHub-native issue and pull request work.
- [Grok Build](https://devthrottle.com/coding-agents/grok-build): xAI's agentic coding CLI that drops the Grok models into the terminal with plan mode, parallel subagents, native MCP, and a headless CI mode, plus a one-line Windows PowerShell installer.
- [Kiro CLI](https://devthrottle.com/coding-agents/kiro-cli): The Kiro terminal agent for writing, reviewing, modifying code, and automating workflows from the command line; AWS positions it as the successor to Amazon Q Developer CLI.
- [Kimi Code CLI](https://devthrottle.com/coding-agents/kimi-code): Moonshot AI's open-source terminal coding agent with local file edits, shell commands, MCP configuration, skills/plugins, subagents, lifecycle hooks, ACP editor integration, `kimi -p` prompt mode, and Agent SDK automation.
- [OpenCode](https://devthrottle.com/coding-agents/opencode): An open-source terminal-based agentic coding tool from the SST team, designed for multi-provider LLM support and a keyboard-driven TUI workflow.

## Other CLI-capable agents

- [Amazon Q Developer CLI](https://devthrottle.com/coding-agents/amazon-q-cli): The former AWS command-line coding agent. AWS documentation now says Amazon Q Developer CLI has been rebranded to Kiro CLI, so this entry is retained for lineage and migration context.
- [Auggie CLI](https://devthrottle.com/coding-agents/auggie-cli): Augment Code's terminal agent for interactive and automated workflows, with codebase indexing, edits, PR/build feedback, MCP server mode, ACP integration, and script-friendly output.
- [Browser Harness](https://devthrottle.com/coding-agents/browser-harness): Browser Use's open-source, self-healing CDP harness that connects a coding agent directly to the user's real running Chrome, with coordinate clicks that pass through iframes, shadow DOM, and cross-origin boundaries at the compositor level.
- [Codebuff](https://devthrottle.com/coding-agents/codebuff): A Y Combinator-backed multi-agent CLI coding tool with a free tier (Freebuff), using specialised agents for file picking, planning, editing, and reviewing to autonomously modify codebases.
- [Cline CLI](https://devthrottle.com/coding-agents/cline-cli): Cline's CLI and SDK surface for bringing its IDE-origin agent into terminal and headless workflows, including MCP, plugins, hooks, skills, rules, connectors, and NDJSON output.
- [Crush](https://devthrottle.com/coding-agents/crush): A Go-based TUI terminal coding agent from the makers of Glow and Bubble Tea, with LSP integration, MCP server support, session management, and a --yolo mode for fully autonomous operation.
- [Devin CLI](https://devthrottle.com/coding-agents/devin-cli): Cognition's local terminal agent for Devin workflows, with file mentions, image paste, shell command execution, `devin -p` script mode, MCP, skills, hooks, AGENTS.md, and cloud handoff.
- [Factory Droid CLI](https://devthrottle.com/coding-agents/factory-droid-cli): Factory's Droid terminal interface for interactive REPL sessions, `droid exec` headless tasks, session resume/fork, daemon workflows, MCP, skills, hooks, custom droids, and enterprise SDLC automation.
- [ForgeCode](https://devthrottle.com/coding-agents/forgecode): A terminal-based AI pair programmer with a three-agent architecture — implement, research, and plan — supporting 300+ models and offering interactive TUI, one-shot CLI, and shell plugin modes.
- [Gemini CLI](https://devthrottle.com/coding-agents/gemini-cli): Google's open-source Gemini terminal agent. Retained for open-source and migration context because Google moved free, Pro, and Ultra individual usage to Antigravity CLI on June 18, 2026.
- [Goose](https://devthrottle.com/coding-agents/goose): Block's open-source agentic developer tool that runs locally, supports any LLM provider, and extends its capabilities through a rich MCP and custom-extension ecosystem.
- [gptme](https://devthrottle.com/coding-agents/gptme): One of the earliest open-source terminal coding agents (Spring 2023), combining code execution, file editing, web browsing via Playwright, and shell access in a persistent conversation interface.
- [Letta Code](https://devthrottle.com/coding-agents/letta-code): Letta's open-source local coding-agent harness with persistent agent state, headless text/json/stream-json modes, MCP, hooks, skills, subagents, and trusted local mods.
- [mini-SWE-agent](https://devthrottle.com/coding-agents/mini-swe-agent): A radically minimal autonomous software engineering agent — approximately 100 lines of core Python — that scores above 74% on SWE-bench Verified using only bash commands in a sandboxed environment.
- [Mistral Vibe](https://devthrottle.com/coding-agents/mistral-vibe): Mistral AI's open-source Vibe Code CLI with interactive and non-interactive `vibe --prompt` workflows, JSON/streaming output, MCP, skills, custom agents, hooks, ACP, VS Code, and web surfaces.
- [Open Interpreter](https://devthrottle.com/coding-agents/open-interpreter): An open-source terminal interface that lets language models run code locally, giving them access to your file system, browser, and any application on your computer.
- [PI](https://devthrottle.com/coding-agents/pi): A minimalist open-source terminal coding agent that runs across 15+ LLM providers, with a TypeScript-based Skills system for extensibility and session branching for non-destructive exploration.
- [Plandex](https://devthrottle.com/coding-agents/plandex): An open-source terminal coding agent designed for large, multi-file development tasks with a cumulative diff sandbox that stages all AI-proposed changes for human review before applying them.
- [Qwen Code](https://devthrottle.com/coding-agents/qwen-code): Alibaba's open-source agentic CLI coding tool, forked from Gemini CLI and optimised for Qwen3-Coder models, with full support for other major providers and VS Code and Zed integrations.
- [SWE-agent](https://devthrottle.com/coding-agents/swe-agent): A research-grade autonomous CLI agent from Princeton and Stanford that takes a GitHub issue URL and resolves it end-to-end using a custom Agent-Computer Interface optimised for software engineering.
- [Trae Agent](https://devthrottle.com/coding-agents/trae-agent): ByteDance Software Engineering Lab's open-source CLI coding agent that achieved 75.2% on SWE-bench Verified using a multi-model ensemble with Claude, Gemini, and o4-mini.
- [Warp](https://devthrottle.com/coding-agents/warp): An AI-native terminal emulator that went open source in May 2026, combining a high-performance Rust-based shell with built-in AI coding agents and the Oz cloud orchestration platform for running autonomous background agents.

## Hosted, browser, and review-agent pages

- [Bolt](https://devthrottle.com/coding-agents/bolt): StackBlitz's browser-based agentic development environment using WebContainers to generate, run, and deploy full-stack applications entirely in the browser with no local installation required.
- [CodeRabbit](https://devthrottle.com/coding-agents/coderabbit): An automated code review agent that triggers on every pull request, generating plain-English walkthroughs, sequence diagrams, and line-level bug and security findings across GitHub, GitLab, Azure DevOps, and Bitbucket.
- [Devin](https://devthrottle.com/coding-agents/devin): An autonomous AI software engineer that accepts a task, spins up its own sandboxed environment, and works end-to-end on coding tasks with minimal human checkpointing.
- [Factory](https://devthrottle.com/coding-agents/factory): An enterprise autonomous coding platform where specialised AI Droids handle the complete software development lifecycle — feature development, code review, incident response, migrations, and documentation — integrated with GitHub, Jira, Slack, and PagerDuty.
- [GitHub Copilot Coding Agent](https://devthrottle.com/coding-agents/copilot-coding-agent): GitHub's cloud-based autonomous coding agent that accepts an assigned GitHub issue, researches the repository, writes code on a branch, runs tests via GitHub Actions, and opens a draft pull request — with no developer involvement during execution.
- [Greptile](https://devthrottle.com/coding-agents/greptile): An AI code review platform that indexes the entire codebase into a call-graph and deploys a swarm of parallel agents per pull request, catching cross-file dependency issues that diff-only reviewers miss.
- [Jules](https://devthrottle.com/coding-agents/jules): Google's cloud-based autonomous coding agent that accepts a GitHub issue, creates a plan, executes it in an isolated Google Cloud VM powered by Gemini 2.5 Pro, and opens a pull request — entirely without developer involvement.
- [Lovable](https://devthrottle.com/coding-agents/lovable): A browser-based AI full-stack app builder that generates, iterates on, and deploys React/TypeScript applications from natural-language prompts, with GitHub integration, Supabase backend, and an autonomous Agent Mode.
- [OpenHands](https://devthrottle.com/coding-agents/openhands): Open-source autonomous engineering platform with real `openhands` CLI, headless CI/scripting mode, MCP support, and web/TUI/server/ACP surfaces for sandboxed issue-to-PR work.
- [Potpie](https://devthrottle.com/coding-agents/potpie): Open-source CLI-first context graph and agent harness layer that provisions local storage, daemon, graph explorer, and integrations with coding harnesses like Claude Code, Codex, Cursor, and OpenCode.
- [Replit Agent](https://devthrottle.com/coding-agents/replit-agent): Replit's cloud-based AI development agent that autonomously builds, deploys, and maintains full-stack applications from natural-language descriptions, running in Replit's cloud IDE with 160+ service connectors.
- [Rovo Dev](https://devthrottle.com/coding-agents/rovo-dev): Atlassian's agentic coding platform deeply integrated with Jira, Confluence, and Bitbucket — converting Jira issues into implementation specs, reviewing PRs against acceptance criteria, and operating via CLI, IDE, and browser with Atlassian's Teamwork Graph as context.

## Notes for crawlers

- Public marketing, docs, blog, and coding-agent pages are allowed in robots.txt.
- App, account, admin, and auth routes are excluded from crawler access.
- Every catalog page is prerendered into static HTML and included in sitemap.xml.