Development Tools

Unsloth Studio is an open-source, no-code web UI for training, running, and exporting LLMs locally. It transforms unstructured files into datasets and lets you fine-tune models 2x faster with 70% less VRAM, all without writing complex training scripts.

Codex now supports subagents, allowing you to spawn specialized, parallel AI workers for complex coding tasks. By defining custom TOML agents with isolated roles (like explorers and reviewers), you can execute multi-step workflows without context rot.

The better Ghostty, Superset, and tmux experience for Windows (and MacOS). Codirigent puts all your sessions in one persistent workspace. - Real-time status per agent - Custom layouts - Git worktree support - Session resume with same permission states. - System notifications like MacOS do - Copy and paste image directly to your session - and more! Built in Rust with GPUI, ~40MB memory, 22MB on disk. Open Source GPL-3.0.

Gemini Embedding 2 is Google's first natively multimodal embedding model that maps text, images, video, audio and documents into a single embedding space, enabling multimodal retrieval and classification across different types of media and it’s available now in public preview.

VENTUNO Q is the first single-board computer born from Qualcomm's acquisition of Arduino. Fusing a 40 TOPS Snapdragon NPU with an STM32 microcontroller creates a unified dual-brain architecture, perfectly balancing heavy AI inference with real-time robotics.

Claude Code now dispatches a team of agents on every PR to catch bugs that skims miss. Available in research preview for Team and Enterprise. It is an AI-powered multi-agent code review that analyzes every pull request like an expert team. It detects bugs, security issues, and hidden logic flaws in AI-generated code, verifies findings to reduce false positives, and delivers high-signal feedback before code reaches production.

OpenClix helps mobile teams run high-impact local push campaigns without heavy setup. Build campaigns with smart triggers, suppression rules, and scheduling, then connect outcomes to retention and engagement metrics. Ship faster, reduce notification fatigue, and iterate campaigns from real performance data.

OpenClix helps mobile teams run high-impact local push campaigns without heavy setup. Build campaigns with smart triggers, suppression rules, and scheduling, then connect outcomes to retention and engagement metrics. Ship faster, reduce notification fatigue, and iterate campaigns from real performance data.

An application security agent that helps you secure your codebase by finding vulnerabilities, validating them, and proposing fixes you can review and patch. Now, teams can focus on the vulnerabilities that matter and ship code faster.

A native Mac app for managing your App Store apps. Edit all locales at once, translate with AI, manage TestFlight, track analytics, reply to reviews. Runs locally – no cloud, no telemetry. Credentials stay in macOS Keychain. Free for one app, Pro unlocks unlimited. Open source.

Step 3.5 Flash is StepFun’s 196B sparse MoE model that activates only 11B parameters per token. It delivers frontier reasoning and strong agentic performance with high efficiency. Seamless native OpenClaw integration makes it one of the best open models for running serious agents right now.

Gemini 3.1 Flash-Lite is the fastest and most cost-efficient model in the Gemini 3 series. At only $0.25 input and $1.50 output per million tokens, it beats 2.5 Flash with 2.5X faster first token and 45% higher output speed while matching or beating quality.

Clean Clode is an open-source text cleaning tool built for developers using Claude Code and OpenAI Codex. Terminal output from Claude Code and Codex frequently includes formatting artifacts like box characters, pipes, irregular wrapping, and excess whitespace. Clean Clode intelligently removes those formatting artifacts while preserving your actual content and structure - so your text is clean, readable, and ready to reuse. Everything runs 100% in your browser. No tracking. No data collection.

Deploy fully autonomous AI agents with a single prompt. BU gives your agents a browser, terminal, and persistent memory - prompt it once, and it keeps running. We've solved authentication and built integrations for Slack, Gmail, Linear, and 100+ more. Supercharged with the best browser agent for monitoring, testing, and scraping the web. Prompt to complex workflow in single API. The future is here.

Type any prompt, get instant shareable links for ChatGPT, Claude, Gemini, Perplexity & Grok — pre-filled and ready to run. Free, private, no sign-up. The fastest Prompt URL generator in 2026.

CommitPlease is a GitHub-based virtual study-with-me for developers. Jump in anytime (no cam/mic), and watch a shared map evolve as you and others commit, PR, and focus. Track your progress, climb the leaderboard, and collect pets along the way.

Mercury 2 ditches sequential decoding for parallel refinement. As the first reasoning diffusion LLM, it generates tokens simultaneously to hit 1,000+ tokens/sec. This delivers reasoning-grade quality inside tight latency budgets for your agentic loops.

GenAI is like a race car without brakes. It accelerates fast — you describe something, and large blocks of code appear instantly. But AI agents silently break things. They remove logic. Relax constraints. Introduce expensive cloud calls. Leak credentials. Change behavior without telling you. git-lrc is your braking system. It hooks into git commit and runs an AI review on every diff before it lands.

Claude Code runs in your terminal. Session data is stored locally, but not in a human-readable format. If the session contains useful context, refactors, tool usage, reviewing or sharing it isn’t straightforward. Claudebin lets you export a Claude Code session. It captures: - the full message thread - file reads/writes - bash commands - web and MCP calls It returns a URL with a structured, navigable viewer. You can link it in a PR, embed a selected range in your docs, or resume it locally.

Synra is a managed MCP gateway. Add your database credentials, get a secure URL, paste it into Claude Desktop — done. No JSON config files, noenv headaches, no local server setup. Currently supports PostgreSQL and Supabase. Read-only by default. Credentials encrypted with AES-256. Built for developers who want to ask their database questions in natural language without spending an hour configuring MCP servers.

An open-weight, native vision-language model built for long-horizon agentic tasks. Its hybrid architecture (linear attention + MoE) delivers the capabilities of a 397B giant with the inference speed of a 17B model.