Muscle-Mem, a behavior cache for AI agents
Hi HN! Erik here from Pig.dev, and today I'd like to share a new project we've just open sourced: Muscle Mem is an SDK that records your agent's tool-calling patterns as it solves tasks, and will deterministically replay those learned trajectories whenever the task is encountered again, falling back to agent mode if edge cases are detected. Like a JIT compiler, for behaviors. At Pig, we built computer-use agents for automating legacy Windows applications (healthcare, lending, manufacturing, etc). A recurring theme we ran into was that businesses already had RPA (pure-software scripts), and…
What it does
In the maker’s words, at launch
Hi HN! Erik here from Pig.dev, and today I'd like to share a new project we've just open sourced: Muscle Mem is an SDK that records your agent's tool-calling patterns as it solves tasks, and will deterministically replay those learned trajectories whenever the task is encountered again, falling back to agent mode if edge cases are detected. Like a JIT compiler, for behaviors. At Pig, we built computer-use agents for automating legacy Windows applications (healthcare, lending, manufacturing, etc). A recurring theme we ran into was that businesses already had RPA (pure-software scripts), and it worked for them in most cases. The pull to agents as an RPA alternative was not to have an infinitely flexible "AI Employees" as tech Twitter/X may want you to think, but simply because their RPA breaks under occasional edge-cases and agents can gracefully handle those cases. Using a pure-agent approach proved to be highly wasteful. Window's accessibility APIs are poor, so you're generally stuck using pure-vision agents, which can run around $40/hr in token costs and take 5x longer than a human to perform a workflow. At this point, you're better off hiring a human. The goal of Muscle-Mem is to get LLMs out of the hot path of repetitive automations, intelligently swapping between script-based execution for repeat cases, and agent-based automations for discovery and self-healing. While inspired by computer-use environments, Muscle Mem is designed to generalize to any automation performing discrete tasks in dynamic environments. It took a great deal of thought to figure out an API that generalizes, which I cover more deeply in this blog: https://erikdunteman.com/blog/muscle-mem/ Check out the repo, consider giving it a star, or dive deeper into the above blog. I look forward to your feedback!
Does the same job
all alternatives →- WRWe're building a desktop app for browser-based AI agents2025 · meha.ai · ▲50
What's up HN! This is Jared and Art. We met on HN and started building together. Over the last few months we've been thinking a lot about how AI agents are going to impact the future. We want agents to be something that's actually useful for normal people as well as the 10x'ers. This lead us to building Meha over the last few months, our first swing at our vision! We saw OpenAI release Operators then we said f*k it let's post. Meha is a desktop app that uses your Chrome browser to execute tasks in the background. It controls your installed Chrome browser and uses LLMs with playwright to plan…
- FAFreeact – A Lightweight Library for Code-Action Based Agents2025 · github.com · ▲122
Hello! We just released freeact (https://github.com/gradion-ai/freeact), a lightweight agent library that empowers language models to act as autonomous agents through executable code actions. By enabling agents to express their actions directly in code rather than through constrained formats like JSON, freeact provides a flexible and powerful approach to solving complex, open-ended problems that require dynamic solution paths. * Supports dynamic installation and utilization of Python packages at runtime * Agents learn from feedback and store successful code actions as…
- UFUse functional tokens for AI agents to simplify app workflows2024 · nexa4ai.com · ▲80
Hi HN! I want to share our latest project at NEXA AI. We developed AI agent foundation models designed to transform how developers create AI agent powered apps. One major challenge we've observed with current human-computer interactions is that many simple, one-step tasks become unnecessarily complex, multi-step workflows due to limitations of current GUIs. AI agents can solve this, but existing AI agent models are slow and costly. To tackle these issues, we built lightweight AI agent models based on our Octopus V2, small language models for function calling (You can learn more about our…
- SDSteel.dev – An open-source browser API for AI agents and apps2024 · github.com · ▲114
- ARA registry of agent benchmarks (including many OSS agent trajectories)2024 · explorer.invariantlabs.ai · ▲6
If you're interested in exploring what LLM-based agent systems these days actually do to solve certain benchmarks such as SWEBench or WebArena, we created a small leaderboard with our team, that allows to view a lot of public and OSS agent results including all the runtime traces (the step-by-step reasoning behind the scenes). Looking at traces is actually quite interesting, as they reveal a lot about the inner working and shortcomings of current agent system, e.g. see https://explorer.invariantlabs.ai/u/invariant/webarena--SteP... for an example trace.
- MOMarinaBox: Open-Source Sandbox Infra for AI Agents2024 · github.com · ▲6
Hey everyone, We're excited to introduce MarinaBox, an open-source toolkit for creating isolated desktop/browser sandboxes tailored for AI agents. Over the past few months, we've worked on various projects involving: 1. AI agents interacting with computers (think Claude computer-use scenarios). 2. Browser automation for AI agents using tools like Playwright and Selenium. 3. Applications that need a live-session view to monitor AI agents' actions, with the ability for human-in-the-loop intervention. What we learned: All these scenarios share a common need for robust infrastructure. So,…
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 16d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, May 2025
the whole month →
- C9
Life & fun · 2025 · felixrieseberg.github.io
