返回首页
最新
https://getreplay.app/<p>Hi builders!
I was falling asleep when the idea for this project came to me. So, I've decided to vibecode it via Lovable to the best of my ability. This is a personal/inspirational project, but I've tried to make it as good as possible given the constraints.<p>Tech Stack:<p>- Frontend: React 18 + Vite + Tailwind CSS<p>- AI: Gemini 2.5 Flash (via Lovable AI Gateway) for low-latency, empathetic coaching (to keep costs low and ship fast)<p>- Infrastructure: Lovable Cloud / Supabase<p>TL;DR: An LLM-powered "What If" view of your potential. Built for fun, inspiration, and research.<p>I would appreciate your feedback, guys. Thank you in advance!<p>P.S. Initial post got bugged, so I had to recreate it without a hyperlink.
This is a declarative orchestration framework for stateless short-lived LLM agents.<p>Agents are orchestrated by hierarchical state machines (HSMs).<p>The HSMs and agents are defined in YAML, with hooks and
webhooks for additional functionality.<p>* FlatAgent: A single LLM call: model + prompts + output schema.<p>* FlatMachine: A state machine that orchestrates multiple agents, actions, and state machines.<p>Examples:
- <a href="https://github.com/memgrafter/flatagents/tree/main/sdk/python/examples" rel="nofollow">https://github.com/memgrafter/flatagents/tree/main/sdk/pytho...</a>
- <a href="https://github.com/memgrafter/research-crawler-flatagents" rel="nofollow">https://github.com/memgrafter/research-crawler-flatagents</a>
- <a href="https://github.com/memgrafter/claude-skills-flatagents" rel="nofollow">https://github.com/memgrafter/claude-skills-flatagents</a>
Raymond here from Butter.dev, an LLM response cache built as a chat-completions proxy. Today we're launching a key feature for the platform: the ability to generalize on dynamic, templated inputs.<p>Caching at the HTTP request level has the obvious problem of generalizability. Nearly no request is identical, due to templated variables (like names) and metadata (like timestamps), so exact-match cache lookups rarely hit. We solve this at Butter by using LLMs to detect dynamic content in requests and derive their inter-relationships, allowing the cache entry to be stored as a template + variables + deterministic code. This allows future requests to contain different variable data, yet still serve from cache.<p>We've found this approach greatly improves cache hit rate, and believe it could be useful for agents performing repetitive back-office tasks, computer use, or data transformations where input data is frequently of the same shape.<p>- You can see a demo of learning patterns here: <a href="https://www.youtube.com/watch?v=ORDfPnk9rCA" rel="nofollow">https://www.youtube.com/watch?v=ORDfPnk9rCA</a><p>- We wrote more about the technical approach here: <a href="https://blog.butter.dev/on-automatic-template-induction-for-response-caching">https://blog.butter.dev/on-automatic-template-induction-for-...</a><p>- It's free to try out here: <a href="https://butter.dev/auth">https://butter.dev/auth</a>