Who it is forA developer who wants a coding agent backed by local or pooled models instead of a paid API.
The problemWiring a coding agent to a self-hosted model means finding the right base URL, model name and flags.
The outcomeYour chosen coding agent talking to the mesh's OpenAI-compatible API.
What you need first
- macOS, Linux or Windows (install script, PowerShell or Homebrew on Apple Silicon)
- One or more machines with a GPU or enough memory for the model
- curl and jq for the README's test requests
Steps
- Start a node with
mesh-llm serve --autoor your private mesh - Launch the agent through the mesh with
mesh-llm opencode,mesh-llm goose,mesh-llm claudeormesh-llm pi - Pick a model the mesh serves and give the agent a small task first
- Read the AGENTS guide in the repo for per agent settings
Prompt for your agent
Start Mesh-LLM/mesh-llm and launch <OpenCode or Goose or Claude Code or Pi> through it, choose the best model the mesh serves for coding, and run a small test task in <repo>.
Caveat. This card is written from the Mesh-LLM/mesh-llm README, not from a test run by Deduped. Before you start, make sure you have macOS, Linux or Windows. Check the repo for newer instructions before you rely on these steps.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for Mesh-LLM/mesh-llm and "Run Goose, OpenCode, Claude Code or Pi against your own mesh"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer