heatdrop.ai

A local compression layer that sits between your agent and the LLM, cutting token counts by routing tool outputs, logs, and code through type-specific compressors.

A skill that turns Claude, Cursor, or any agent host into a social-native research analyst, scoring signals by upvotes, likes, and actual money wagered.

Hermes Agent is a self-improving AI assistant that creates skills from experience, persists knowledge across sessions, and runs anywhere from a $5 VPS to serverless cloud.

Desktop app that corrals Claude Code, Codex, Gemini CLI and half a dozen other AI agents into one switchboard.

A Python pipeline that turns a keyword into a complete short-form video—script, stock footage, voiceover, subtitles, and all.

A developer turned months of manual applications into an AI agent pipeline that evaluates, scores, and tailors CVs for each listing.

Claude Code skills that run your research through a full academic pipeline—research, writing, staged integrity checks, and multi-perspective review—while keeping a human in the driver's seat.

A free, from-scratch curriculum that makes you build backprop before you touch PyTorch, then ships every lesson as a reusable prompt or agent.

23 slash commands and 27 anti-pattern rules that teach Cursor, Claude, and Copilot how to design instead of defaulting to Inter and purple gradients.

Claude Code skill makes the agent talk like a caveman and claims ~65% fewer output tokens, with benchmarks to back it up.

rtk sits between your AI agent and the shell, compressing command output before it ever hits context.

A React SDK that lets AI agents render UI mid-thought and ask humans for help before continuing.

A repo of reusable "skills" that teach agent frameworks how to talk to Google Cloud APIs and follow its well-architected pillars.

Goose is a Rust-built, open-source agent that edits code, runs commands, and connects to 70+ tools via MCP — not just another chat wrapper.

Datawhale's open-source curriculum wants to turn LLM users into agent builders, not just prompt engineers.

whichllm ranks local models by real benchmark scores, not parameter count, and tells you which ones actually fit your hardware.

Garry Tan open-sourced the exact Claude Code prompts he uses to ship 810× faster while running Y Combinator full-time.

Vibe-Trading wraps LLMs in guardrails, mandates, and kill switches before letting them near real money.

A hybrid CLI tool that uses deterministic pipelines to keep LLM agents from drifting off-target during code review.

An open-source gateway for splitting AI subscriptions across teams without breaking native tools.