AI Coding Tools

Terminal agents, inline completion, AI-native editors, local models — and the open-source layer that gives them repository context, specialist roles, and a mission control.

Claude Code

An agentic coding tool that reads a whole repository and edits files from the terminal.

GitHub Copilot

Inline completion inside your editor. Biggest payoff on repetitive code.

Cursor

A VS Code fork that puts AI editing front and centre — and imports your existing setup.

Kiro

AWS's agentic IDE and CLI, built around writing the spec before the code.

Muse Code

Meta's terminal coding agent — persistent subagents in isolated worktrees, rewind on double-Esc, a single static binary — and a default …

llmfit

One command that reads your RAM, GPU, and VRAM and tells you which local models will actually run — before you download 20 GB to find out.

Ollama

Running models on your own machine, and what to realistically expect from them.

Slotstream

Running a 105 GB mixture-of-experts model on a Mac that doesn't have 105 GB of RAM, by streaming experts from SSD.

edge0

The same SSD-streaming trick as Slotstream, aimed the other way — a 35B model held to roughly 1–3 GB of live memory, with a router that …

Open WebUI

The ChatGPT-shaped front end for models you host yourself — and for every API key you already have.

FreeLLMAPI

Stacking thirty-four providers' free tiers behind one OpenAI-compatible endpoint — with the Docker and Podman startup that actually works.

Needle

An 8–29 MB foundation model that does tool calls, typed extraction and embeddings on phones, robots and microcontrollers — with a confidence …

Freebuff

A coding agent with no subscription, no credits, and no API key — funded by ads, which is the part to read carefully.

Luvus

A terminal mission control that runs, watches, resumes, and orchestrates several AI coding agents in one place.

Orca

A desktop IDE that gives each agent task its own git worktree, terminal, browser tab, and diff to review.

herdr

tmux rebuilt for the era of several coding agents at once — every pane marked working, blocked or idle, and a socket API the agents …

Graft

A pre-built map of your repository so the agent stops re-reading the same files on every task.

Codebase Memory MCP

Indexing a repository into a queryable graph an agent can ask structural questions of, instead of grepping.

Agency Agents

A library of 230+ ready-made subagent definitions across seventeen divisions — and why you should install about six of them.

Agentic SDLC Kit

Twelve Agent Skills that carry a project from an interview to a verified, deployed system — with stable IDs, a shared traceability matrix, …

Frontend Design Skills

Five skills that give an agent taste — aesthetic direction, a UI audit, motion fixes, and a design-decision library.

book-to-skill

Turning a technical book, a docs folder, or a stack of papers into a skill your agent loads a chapter of when it needs one.

OpenMontage

Turning a coding agent into a video production pipeline — research, script, assets, edit, render — from one sentence.

HyperFrames

Writing a video as HTML, CSS, and a paused timeline — a rendering framework built so an agent can author the file.

Remotion

Making real MP4 files out of React components — the framework HyperFrames is a reaction to.

Composio

Giving an agent authenticated access to real applications — Gmail, Slack, GitHub, Jira — without building an OAuth flow per service.

Exa

Web search built for models rather than people — meaning-based retrieval, clean extracted text, and a research endpoint with citations.

Browser Use

An agent that opens pages, clicks, types, and fills forms — the library most of the AI-browsing world is built on.

Camofox Browser

A browser server for agents: accessibility snapshots instead of HTML, stable element refs, and a Firefox build that ordinary bot detection …

Vibe-Trading

An open-source research workspace where finance questions become runnable backtests — 90 skills, agent teams, and an audit trail.

AutoHedge

A four-agent pipeline — thesis, quant, risk, execution — that trades a Solana wallet on its own.

promptfoo

Running your prompts as a test matrix from the command line, then pointing 50+ attack categories at the result — MIT licensed, local, and …

DeepEval

Pytest for LLM applications — 50+ research-backed metrics, thresholds that fail the build, and a judge you can point at your own criteria.

Langfuse

Self-hosted tracing and evaluation for LLM apps — every call, cost and score on a timeline you own, MIT licensed and uncapped.

Paperclip

An org chart, ticket system, and budget ceiling for a fleet of AI agents — the control plane for when twenty Claude Code tabs stop being …