AI repos for full-stack developers
Open-source AI projects worth it for people who build whole web apps, front to back, each with what it does for you. 65 repos, most relevant first.
-
claude-code
Terminal agent that explains code, makes edits and handles git workflows across a whole app.
-
playwright
One API for end-to-end testing and browser automation of whole web apps.
-
ai
Add streaming chatbots and Zod-validated structured output to Next.js apps.
-
opencode
Open source terminal coding agent with read-only plan mode and build mode for everyday development.
-
gstack
Slash commands for code review, browser QA and release steps in Claude Code.
-
codex
Terminal coding agent powered by a ChatGPT plan or API key to read and change project code.
-
awesome-claude-code
Browse guides, skills and status lines to get more out of Claude Code in daily development.
-
agent-skills
Gives agents Vercel deployment and Next.js rules, and finds caching and function cost problems.
-
page-agent
Add an AI copilot to an existing SaaS app without rewriting the backend.
-
roomGPT
See how a Next.js API route calls a Replicate ControlNet model, then fork it for another image-to-image app.
-
worktrunk
Review PRs in a separate worktree and run parallel agent branches without disturbing current work.
-
improve
Get a ranked list of problems with file and line evidence and turn them into implementable specs.
-
no-mistakes
Gate AI-written branches through review, tests and lint in a throwaway worktree before a PR opens.
-
proof-sdk
Build collaborative markdown editing with realtime server, edit history and an agent HTTP bridge.
-
vibe-coding-playbook
Cursor rules, plan-first XML prompts and a pre-commit verify script for AI-written TypeScript.
-
claude-code-mastery-project-starter-kit
Scaffold Claude Code projects with slash commands, hooks and skills for review, testing and commit guards.
-
aura-code
Use a terminal coding agent with persistent memory to refactor and verify code with a model you choose.
-
Logic-Loop
Run parallel Claude Code or Codex sessions across projects and track open questions and blockers.
-
claude-dashboard
Monitor several Claude Code terminals, review file diffs per session, and track costs against a budget.
-
superpowers
Makes coding agents plan, write tests first and review before shipping multi-task features.
-
skills
Slash commands that grill requirements, drive test-first builds and diagnose bugs in existing projects.
-
ECC
Adds plan, test, review and language-specific rules to Claude Code across a project.
-
andrej-karpathy-skills
Drop-in CLAUDE.md that cuts bloated diffs and silent guesses from AI-generated code.
-
ponytail
Rule set that stops agents adding needless libraries and reviews existing projects for over-built code.
-
open-webui
Study a full self-hosted chat app with user roles, document search and a plugin system.
-
pi
Customizable terminal coding agent you can script in print or JSON mode.
-
caveman
Lower your daily coding agent bill with shorter answers and compressed test output.
-
agent-skills
Skills and slash commands make coding agents follow spec, plan, test-driven build, review and ship steps.
-
paperclip
Node.js server plus React dashboard for tracking agent tasks and spend, useful to run and study.
-
orca
Run several coding agents in parallel git worktrees and review their diffs in one app.
-
claude-mem
Stops re-explaining the project to Claude Code each session and lets you search past bugfixes.
-
worldmonitor
Self-hostable web and desktop dashboard with data aggregation and an MCP server to study.
-
archify
Turn a codebase or description into explorable architecture and data-flow diagrams.
-
Understand-Anything
Knowledge graph of files, functions and dependencies for getting oriented in an unfamiliar codebase.
-
screenshot-to-code
Get UI scaffolds from images and self-host the tool with your own API keys.
-
ruflo
Run multi-agent teams on larger coding tasks, with test generation and vulnerability scanning plugins.
-
warp
Use one terminal for shell work and agents like Claude Code, Codex or Gemini CLI.
-
multica
Hand bug reports to Claude Code or Codex and review the resulting pull requests in one place.
-
open-code-review
Get automated line-level review of Git diffs before opening pull requests.
-
codebase-memory-mcp
Ask coding agents who calls a function and what a refactor affects, without reading many files.
-
herdr
Run several coding agents side by side and see which one is blocked or waiting.
-
playwright-mcp
Have an AI assistant exercise web app flows and write tests with Playwright.
-
postiz-app
Self-hostable Docker, Postgres and Redis app with an API and MCP for automation.
-
ai-website-cloner-template
Migrate your own sites from WordPress or Webflow into a Next.js codebase using an AI agent.
-
skills
Share team skills across coding agents by committing them to a project.
-
claude-code-templates
Installs ready-made agents, commands, hooks and MCP configs into Claude Code with one npx command.
-
context-mode
Keeps long coding sessions on track by cutting context from large logs and Playwright snapshots.
-
supermemory
Adds persistent user memory to chatbots and apps via a hosted API or local server.
-
agentmemory
Keeps project context and past decisions across coding agent sessions so you stop re-explaining.
-
t3code
Switch between coding agents and steer long-running tasks remotely from one UI.
-
cmux
Manage parallel Claude Code or Codex sessions with notifications and a built-in browser pane.
-
compound-engineering-plugin
Turn feature ideas into plans and multi-agent reviewed code with a fixed command sequence.
-
ralph
Turns a written feature spec into tested commits on a branch with little hand-holding.
-
dyad
Local alternative to Lovable or v0 for prototyping web apps with your own API keys.
-
openui
Add generative UI to a chat product with React and LangChain/LangGraph streaming.
-
inbox-zero
Real TypeScript app with AI rules, email integrations and self-hosting to read and modify.
-
sandcastle
Run several coding agents in parallel on separate branches in TypeScript while away from the keyboard.
-
openrig
Run a builder and reviewer agent pair on a codebase from one command and watch them in a terminal dashboard.
-
eve
Add Slack or Discord agents and typed tools to a Node and TypeScript project with a file-based layout.
-
skills
Skills for explaining unfamiliar code with diagrams and writing PRs with visual outlines for reviewers.
-
llms-txt
Add an llms.txt and .md page variants to a site or docs build.
-
OpenTag
Starter app showing Slack app setup, Node plumbing and a Python agent backend together.
-
Iron-Jarvis
Delegate coding tasks to subagents and review git diffs before merging, all on a local machine.
-
screencontext
Record a reproduced bug and hand the video to an agent instead of describing it.
-
notch-agent-hud
See which Claude Code and Codex sessions are waiting, and get warned before hitting quota limits.