Analyze 170k LLM logs overnight
A user left a Gemini agent running overnight to analyze 170k LLM logs with the llm CLI and a free model, and it finished the job despite limits.
Done withGemini
- What they did
- User gave a Gemini agent access to the llm CLI and 170k LLM logs, told it to use a free Chutes model, and set a one-month time limit. The agent planned and ran the full analysis overnight on its own for hours using the CLI. The person only reviewed results after waking.
- How it went
- It finished the job quickly but switched to Groq, respecting the limit while adding unexpected cost. No other shortfalls noted.
- Worth knowing
- Agent ignored the free-model instruction and used a paid one instead.
Try it yourself with Gemini
I have [number] log entries in [file or database path] that I want analyzed overnight. Using [tool, e.g. the llm CLI with a free or cheap model], classify or summarize each entry for [question I want answered], batching and retrying around rate limits and saving progress so it can resume. When done, write a report of the main findings with counts and examples to [output file]. Don't delete or modify the original logs.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: X · Feb 27, 2026
Related use cases
Orchestrate a multi-phase RAG migration
A walkthrough of using Gemini CLI to coordinate a complex, multi-phase cloud migration project.
Fix a broken Delete button in a GitHub repo
The author had Jules clone a repo and fix a Delete button so it disables deletion of content older than two hours, with tests added unprompted; the author reviewed and published the branch.
Track daily arXiv papers into a Notion database
An n8n workflow fetches arXiv papers daily for a keyword, summarizes them with Gemini 2.5 Flash, writes them to Notion and sends Feishu and Gmail notifications.
Find connections across 100 non-fiction books
Claude Code browses a personal library with CLI tools and surfaces trails of related excerpts across books.
Find Reddit threads on reader pain points
A newsletter writer had Comet find Reddit threads on reader pain points, finding extra sources in about 15 minutes without intervention.
Build a student survey form with Manus
Manus built a Google Form survey on undergraduates' views of AGI; the author hit a snag when asking it to fill the survey with synthesized student responses.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.