lmstudio-log-observability

name: lmstudio-log-observability description: "Stream, capture, and inspect LM Studio local server, runtime, and model I/O logs. WHEN: "lms log", "LM Studio logs", "debug local LLM server", "capture model input output", "server log stream"." license: MIT metadata: author: ChatGPT version: "1.0.0"

LM Studio Log Observability

Use this skill to debug LM Studio local server behavior with repeatable lms log stream workflows.

Workflow

1. Verify CLI access

   lms --help
   lms log stream --help
   

2. Start or confirm the local server

   lms server start
   lms server status
   

3. Choose the log source

   Goal	Command
   Stream HTTP/server events	lms log stream --source server
   Short form	lms log stream -s server
   Inspect prompt/model I/O	lms log stream --source model --filter input,output
   Save server logs	lms log stream -s server | tee -a lmstudio-server.log
   Save JSONL logs	lms log stream -s server --json | tee -a lmstudio-server.jsonl

4. Generate traffic from another terminal, client, or test harness.

5. Correlate failures by timestamp, endpoint, model identifier, response status, and whether the issue appears in server logs, model I/O, or runtime behavior.

6. Use installed CLI help as authority when sources differ. Some installs may expose extra sources such as runtime; only use them if lms log stream --help lists them.

When More Detail Is Needed

• Use command recipes for capture, filtering, and replay commands.
• Use troubleshooting for common failure patterns.
• Use automation scripts when the user wants reusable capture commands.

Output Pattern

When responding, provide:

1. The exact log command to run.
2. The companion command that generates server traffic.
3. The expected signal in the logs.
4. A minimal next diagnostic step if no logs appear.