A developer shipped major updates to the open-source vscode-localllm extension, which lets users run local Ollama LLMs inside VS Code agent chat. The update adds MCP and tool-skill support, an activity tracker with usage telemetry, and fixes to the git workflow. It targets practitioners who want local model use in coding agents without sending code to cloud APIs.
A developer launched 'Local Ollama Chat', a VS Code extension connecting the editor's agent chat to a local Ollama server, citing lag and JSON-garbled output in the existing Ollama extension. It reads workspace files and proposes file edits as reviewable diffs, keeping code entirely local. The project is open source and the author is soliciting feedback.
A developer launched 'Local Ollama Chat', a VS Code extension connecting the editor's agent chat to a local Ollama server, citing lag and JSON-garbled output in the existing Ollama extension. It reads workspace files and proposes file edits as reviewable diffs, keeping code entirely local. The project is open source and the author is soliciting feedback.
Hillnote is a local-first Markdown workspace that integrates AI capabilities through a CLI, using Ollama and MCP. It solves the problem of having a private, offline-friendly note-taking and document editing environment with AI assistance, targeting users who want to keep their data local while leveraging AI models.
Hillnote is a local-first Markdown workspace that integrates AI capabilities through a CLI, using Ollama and MCP. It solves the problem of having a private, offline-friendly note-taking and document editing environment with AI assistance, targeting users who want to keep their data local while leveraging AI models.
Selfcoder is a VS Code coding agent extension built to work with local models through Ollama. It manages context deliberately, selecting relevant workspace context so smaller local models get what they need without wasting their limited context windows. It targets developers frustrated with poor context handling when using local models via native Copilot Chat.
Selfcoder is a VS Code coding agent extension built to work with local models through Ollama. It manages context deliberately, selecting relevant workspace context so smaller local models get what they need without wasting their limited context windows. It targets developers frustrated with poor context handling when using local models via native Copilot Chat.
A developer published a setup guide for running Gemma 4 or Qwen 3.5 as a local coding agent on a 32 GB Mac. The guide covers Ollama vs LM Studio, model selection for 32 GB, OpenCode with local inference, and MCP persistent memory for cross-session context. It includes OAuth flow for MCP auth and notes on Gemma 4's thinking mode for debugging.
A developer published a setup guide for running Gemma 4 or Qwen 3.5 as a local coding agent on a 32 GB Mac. The guide covers Ollama vs LM Studio, model selection for 32 GB, OpenCode with local inference, and MCP persistent memory for cross-session context. It includes OAuth flow for MCP auth and notes on Gemma 4's thinking mode for debugging.
xNode360 is a live map of your infrastructure where every node self-reports status and latency. It now ships as a VS Code extension with a side-panel map and status-bar failure count, plus an MCP server so LLM assistants can query the live infra state. It targets developers whose setups have grown too complicated to track mentally.
xNode360 is a live map of your infrastructure where every node self-reports status and latency. It now ships as a VS Code extension with a side-panel map and status-bar failure count, plus an MCP server so LLM assistants can query the live infra state. It targets developers whose setups have grown too complicated to track mentally.