docs: simplify sharing Ollama, vLLM and LM Studio through plugins - #1823
Conversation
Document provider setup, private-mesh consumption, release compatibility and the HTTP/authentication limits without adding another runtime command. Signed-off-by: Michael Neale <14976+michaelneale@users.noreply.github.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Advanced Run ID: 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (1)
Included review availability: Your plan provides up to 8 included reviews per hour; 7 remain after this review. 📝 WalkthroughWalkthroughThe documentation updates external endpoint setup, plugin configuration, adapter compatibility, navigation, FAQ entries, and troubleshooting guidance for Ollama, vLLM, LM Studio, and other OpenAI-compatible providers. ChangesExternal model endpoint support
Priority: ⬇️ Low Estimated code review effort: 2 (Simple) | ~10 minutes Change: Other 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
i386
left a comment
There was a problem hiding this comment.
Reviewed the docs diff in full — clear provider recipes, honest limitations (no direct https upstream, adapter control connection semantics), consistent nav/data updates. Second approval alongside i386; green CI.
Signed-off-by: Michael Neale <14976+michaelneale@users.noreply.github.com>
Summary
Make sharing an existing Ollama, vLLM or LM Studio server a short, explicit plugin workflow rather than a second runtime command:
Start the provider → install a compatible
openai-endpoint→ set its URL →mesh-llm serve.client --joinfrom a second machine.Companion and rollout
Adapter repair: Mesh-LLM/openai-endpoint#3
Prefer merging the adapter first. Guide links to its compatibility/build instructions are pinned to the reviewed commit so they also work before merge. The adapter's source-build recipe selects its rescue branch; retain it until replaced with a durable release/main reference. Publishing a compatible adapter release is a separate required step—this documentation does not pretend catalog installs are repaired already.
Published adapter 0.1.2 speaks protocol 2 and cannot initialize on Mesh 0.76.0 (protocol 3). Older protocol-2 hosts need their compatible pin. No host code, protocol behavior, or runtime command changes in this PR.
Example
For a running local Ollama server, edit
~/.mesh-llm/config.toml:Start
mesh-llm serve, listhttp://127.0.0.1:9337/v1/models, and send a completion using an exact listed ID. Change only the URL for vLLM (:8000/v1) or LM Studio (:1234/v1).Validation
npm run build: CLI inventory, Tailwind, Eleventy, Pagefind passed.npm run test:cli-explorerpassed.node --check website/src/_data/docs.jspassed.git diff --checkpassed. Generated output is not committed; Rustdoc regeneration/full Mesh runtime build was not needed for these authored docs/navigation changes.Limits
Recipes checked against provider documentation and host/adapter source, not live certification of each provider. Prior adapter validation used two official Mesh 0.76.0 processes on one Mac with an HTTP fixture, not two machines or real-model/streaming/auth recovery testing. Direct HTTPS upstream and upstream API-key injection are not supported by this URL-only recipe; docs keep unauthenticated providers loopback-only.
Summary by CodeRabbit