Mini Me agents can talk to each other over #ACP. So I built a team: a coordinator (Qwen3.5 0.8B) that hands my Go and Rust questions to two specialists sharing the same Mellum2 model. Interactive docs, all local 🤗
@k33gorg.bsky.social
@k33gorg.bsky.socialMini Me agents can talk to each other over #ACP. So I built a team: a coordinator (Qwen3.5 0.8B) that hands my Go and Rust questions to two specialists sharing the same Mellum2 model. Interactive docs, all local 🤗
@k33gorg.bsky.socialNew #sbx kit: Mini Me Editor. It bundles the Mini Me agent and the #Eclipse #Theia web IDE in a single @docker.com sandbox, so I get a fully isolated dev environment in my browser. A bit like GitPod 🍊, but on my machine 🤗
@k33gorg.bsky.social#ACP is to coding agents what #LSP is to language servers: one protocol, any compatible editor.
Mini Me speaks it 🙂 Here is how to use it from #Zed and #VSCode (with ACP Patchbay), directly or from inside a @docker.com sbx sandbox.
mini-me.org/p/20261011-m...
#LocalModels
@k33gorg.bsky.socialRunning an AI agent directly on your machine? It sees your configs, ssh keys, env variables, secrets...
So Mini Me now has its @docker.com sbx kit: two files and one command to run it in a sandbox. You'll see, it's quick.
mini-me.org/p/20261010-m...
#sbx #LocalModels
@k33gorg.bsky.social@devoxx.com Belgium reboosts your brain, so my baby Mini Me got a few new features during these 5 days 🙂
v0.3.0: approve tool calls or not (yolo: false), agents calling other agents, request traces, JSON answers, AGENTS.md...
mini-me.org/p/20261010-m...
#LocalModels
@trbutler.faithtree.socialI hope a nice, quantitized version of this shows up so I can try it with #MLX #oMLX.
#AI #LocalModels #DataSovereignty https://flipboard.com/@cnet/tech-tips-%F0%9F%93%B1%F0%9F%92%BB%F0%9F%93%A1-a20abmejz/-/a-R7H13fpdRD2YWFm63mDbgg%3Aa%3A24934323-%2F0
@nile11.bsky.socialGoogle AI Edge Foresight Brings Offline Meeting Notes to Mac
..................
https://nile1.com/google-ai-edge-foresight-brings-offline-meeting-notes-to-mac/
..................
#aiNotetaker #appleSilicon #embeddinggemma2 #fathom #gemma4 #googleAiEdgeForesight #granola #localModels #offlineMeetingN
@eicker.bsky.social#Ghost, founded by #ZainJavaid and three others, raised $11 million from #AndreessenHorowitz and others for #Core, a $3,499 #AIcomputer. Core runs #localmodels and #encrypteddata, with preorders opening Monday and shipping in October.…
@k33gorg.bsky.socialPreparing it, small local models kept failing on me: hallucinations, loops, truncated tool calls...
So a side project was born: Mini Me, a coding agent built for small local LLMs, with guardrails against exactly that.
mini-me.org/p/20261003-h...
#LocalModels
@k33gorg.bsky.socialPlugged #Zed's agent panel into @docker.com Agent over #ACP, then pointed it at #llmman: a coding agent that reads files, runs shell commands and answers me, entirely on my own machine. No cloud API involved.
@k33gorg.bsky.socialllmman: "Run any agent on any model". It pulls models like OCI images and serves them behind one API that speaks OpenAI, Ollama and Anthropic.
I tried it with llama.cpp as the runtime, on macOS and Linux, plus a Go example.
@wagesj45.mastodon.jordanwages.com.ap.brid.gyBecause fuck Dario Amadeus Mozart
@eicker.bsky.socialThe #M5Ultra #MacStudio is a powerful machine for running local #AIagents, outperforming its predecessor and even a high-end gaming PC in some areas. With its improved #GPU and #memorybandwidth, it enables faster #promptprocessing and #generation, making #localmodels a viable option for tasks like…
@wideareaai.bsky.social@wideareaai.bsky.social@wideareaai.bsky.social@wideareaai.bsky.social@k33gorg.bsky.socialWhen a new model shows up in GGUF, #llama.cpp is usually the first to handle its specifics. So that is what I point my code agent at for testing new models.
`llama serve`, a short `agent.yaml`, and @docker.com Agent runs fully local, shell and files included.
@wideareaai.bsky.social