alphacode
Free MIT AI coding agent — no API key needed. Built-in free model, or bring Claude, GPT, Gemini, DeepSeek, Ollama +50 more. Swarm mode, 40+ tools, browser & desktop automation, built in Rust.
Install / Use
npx skills add dragonked2/alphacodeInstalls into whichever agent you are using.
Other
Other agent config
Quality Score
Category
AutomationSupported Platforms
Our assessment of alphacode
alphacode scores 82/100 on our quality scale, 2339th of 2,869 Automation skills we index.
Its Other is 34 KB long, well organised into 60 sections with 41 code examples: a thorough specification that gives an agent plenty to work with.
It has 211 GitHub stars, a meaningful sign that others use it.
Maintenance, license and trust
- The repository was last updated yesterday, so alphacode is actively maintained.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
alphacode compared with similar skills
All 4 of these similar skills score higher than alphacode; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| alphacode (this skill)by dragonked2 | 82 | 211 | 1d ago | Other |
| Agent-Reachby Panniantong | 100 | 91.8k | 20d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.5k | today | CLAUDE.md |
| Scraplingby D4Vinci | 100 | 85.9k | 1d ago | MCP Server |
| crawl4aiby unclecode | 100 | 84.8k | today | MCP Server |
Frequently asked questions
- How do I install alphacode?
- Run
npx skills add dragonked2/alphacode. The install tabs above show the steps for each supported agent. - Which AI agents does alphacode work with?
- It is written for Claude Code and Gemini CLI, as a Other file. Other agents that read the same format can often use it too.
- Is alphacode safe to use?
- It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is alphacode still maintained?
- The repository was last updated yesterday, so alphacode is actively maintained.
Skill content
View source on GitHubAlphaCode: the open-source AI coding agent that builds, browses, automates and verifies
A free, Rust-native AI coding agent for your terminal. It reads your codebase, edits files, runs commands and tests, drives a real Firefox browser, controls desktop apps, and verifies its own work.
What is AlphaCode? · Quick Start · Features · Browser Agent · Desktop Control · Swarm Mode · Providers · Benchmarks · Uninstall · FAQ
</div>What is AlphaCode?
AlphaCode is an open-source AI coding agent that works inside your development environment instead of just chatting about it. Describe a goal in plain English and it investigates, plans, edits files, runs commands, uses the web, operates a real browser, interacts with desktop apps, runs your tests, and reports back with evidence.
flowchart LR
A["You<br/>describe a goal"] --> B["Understand<br/>project + context"]
B --> C["Plan<br/>choose actions"]
C --> D["Act<br/>code, browser, desktop"]
D --> E["Verify<br/>tests + evidence"]
E --> F["Report<br/>what changed"]
Most AI assistants stop at You → AI → text. AlphaCode is built around a loop: understand → plan → execute → observe → verify.
Example tasks:
Find why the app crashes when users upload a PDF.
Fix the root cause, add a regression test, run the test suite, and review the final diff.
Open the site in Firefox, log in with my existing session,
find the account settings page, check that the new feature is visible,
and send me a screenshot.
Audit this project for security problems. Trace untrusted input to sensitive operations,
validate the realistic findings, and tell me which ones are actually reproducible.
Who it is for: beginners can use plain English, developers can give precise technical instructions, and power users can orchestrate multiple agents, skills and MCP servers.
Quick Start
1. Install
macOS / Linux
curl -fsSL https://raw.githubusercontent.com/dragonked2/alphacode/main/scripts/install.sh | bash
Windows (PowerShell)
irm https://raw.githubusercontent.com/dragonked2/alphacode/main/scripts/install.ps1 | iex
Prefer to read a script before running it? Download it first, inspect it, then run it. See Installation for pinned versions, PATH options, and building from source.
2. Check your setup, then run
alphacode doctor # checks providers, terminal and paths, and tells you what's missing
alphacode # launch the terminal UI
No API key is required to start. The built-in free AI lane works out of the box. To use your own provider, run alphacode login (see Bring your own model). The first launch shows a short onboarding wizard (telemetry choice, model defaults, key bindings), and every step can be skipped with Esc.
3. Give it a task
Explain this project and identify the main entry points.
Find the biggest bugs, fix the highest-impact one, and run the tests.
Or run a single task non-interactively:
alphacode run "fix the failing test"
4. (Optional) Connect Firefox
Install the AlphaCode Browser Agent extension, then:
alphacode browser setup
alphacode browser status
A healthy setup reports that the browser bridge is available and responding.
Features
| Capability | Details | | --- | --- | | Autonomous coding agent | Natural-language tasks, multi-step planning, self-verification | | File and code tools | Read, write, targeted edits, multi-edit, patches, code-aware search | | Terminal | Shell execution with safety controls, background jobs, batching | | Web | Search, fetch, scraping, HTTP flow analysis | | Real Firefox automation | DOM, forms, tabs, frames, screenshots, downloads, existing sessions | | Desktop automation | Windows UI Automation, macOS Accessibility API, Linux AT-SPI2 | | Swarm Mode | Multiple cooperating agents for decomposable work | | Persistent sessions and memory | Resume long-running work across terminal restarts | | MCP support | Connect external tools and services | | Skills | Reusable, specialized workflows | | Multi-provider | Anthropic, OpenAI, Gemini, Copilot, OpenRouter, Bedrock, Azure, self-hosted, and more | | Free AI lane | Start without an API key | | Cross-platform | Windows, macOS, Linux |
Work with your code
Read and create files, apply targeted edits and patches, search repositories, trace code paths, run builds and tests, review diffs, and diagnose errors.
Work with the web
Combine web search and fetching with real browser automation: inspect dynamic pages, interact with web apps, handle forms and browser state, capture screenshots, and assist with QA and security testing.
Browser Agent
The AlphaCode Browser Agent is a Firefox extension that connects AlphaCode to a real Firefox session through native messaging. AlphaCode stays the reasoning and orchestration layer, while Firefox becomes an execution environment.
flowchart LR
U["You"] --> A["AlphaCode agent"]
A --> B["browser tool"]
B --> C["Native messaging"]
C --> D["Browser Agent<br/>Firefox extension"]
D --> F["Real Firefox"]
F --> G["Page: DOM, forms,<br/>frames, UI"]
G --> D
D --> A
A --> R["Verified result"]
A plain HTTP client fetches HTML. A browser agent experiences the page the way a user does, which matters for:
- JavaScript-heavy sites and dynamic interfaces
- DOM inspection, forms, inputs and interactive controls
- Multiple tabs, frames and embedded content
- Existing authenticated browser sessions
- Scrolling, navigation, screenshots and downloads
- Browser-based testing, QA and web-app debugging
- Authorized security testing and repetitive browser workflows
Instead of pasting HTML into a chat and asking "what should I click?", you can say:
Open the application, check the login flow, go to the dashboard,
test the form validation, and report anything suspicious.
Privacy note: the agent can act inside your real browser session. Review what it is asked to do, and treat page content as untrusted input.
Desktop Control
AlphaCode drives native applications through accessibility APIs rather than relying only on screen coordinates.
| Platform | Backend | | --- | --- | | Windows | UI Automation | | macOS | Accessibility API | | Linux | AT-SPI2 |
Supported actions include list_windows, snapshot, find, click, type, press, scroll, focus, screenshot, toggle, expand and collapse.
Open Calculator, calculate 123 × 456, and read the result.
AlphaCode finds the app, inspects its accessibility tree, locates the controls, enters the expression, reads the result, and reports 56088.
Swarm Mode
Large goals can be split across multiple agents working in parallel.
flowchart TD
G["Large goal"] --> P["Planner"]
P --> A["Agent A<br/>code analysis"]
P --> B["Agent B<br/>security review"]
P --> C["Agent C<br/>tests"]
P --> D["Agent D<br/>documentation"]
A --> M["Coordinator"]
B --> M
C --> M
D --> M
M --> R["Review"]
R --> F["Final result"]
/swarm "Analyze this application from architecture, security, testing, and performance perspectives."
Swarm Mode works best on tasks that decompose into independent or semi-independent parts.
Sessions and Memory
Long-running work should survive a closed terminal.
alphacode sessions list # list saved sessions
alphacode --resume # resume the latest session
alphacode --resume <id> # resume a specific session
Useful for large codebases, security assessments, long debugging sessions, multi-stage development, research, browser workflows and big refactors.
Bring Your Own Model
AlphaCode is model-agnostic. Start with the free AI lane, then connect providers when you want to.
| Provider | Auth | Notes |
| --- | --- | --- |
| Anthropic (Claude) | OAuth or API key | First-class support |
| OpenAI (GPT, o-series) | OAuth, API key, ChatGPT browser sign-in | Reasoning, vision and tools |
| Google Gemini | OAuth or API key | |
| GitHub Copilot | OAuth | Uses your existing subscription |
| Cursor | OAuth | Reuses Cursor's session |
| AWS Bedrock | AWS credentials | Build with --features bedrock |
| Azure | Azure AD or API key | Build with --features azure-auth |
| OpenRouter | API key | Aggregator with a free tier |
| Any OpenAI-compatible API | API key | alphacode provider add <name> |
| Free AI lane | Built-in | Works out of the box, no key needed |
alphacode login # interactive provider picker
alphacode login --provider openai # sign in to a specific provider
alphacode provider list # list providers
alphacode provider add <name> # add a custom OpenAI-compatible endpoint
alphacode provider current # show the active provider and model
alphacode model list # list models
alphacode model use <model> # switch model
In the TUI, press Ctrl+T to open the model/provider selector.
Signing in with a provider
Most providers work from an environment variable such as ANTHROPIC_API_KEY or OPENAI_API_KEY. Run alphacode login from your
Truncated for display — read the full file on GitHub.
Related Skills
Agent-Reach
91.8kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
74.5kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Scrapling
85.9k🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
crawl4ai
84.8kOpen-source web crawler and scraper for LLMs and AI agents: any website into clean, LLM-ready Markdown. Run it yourself, or use Crawl4AI Cloud with one key.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
