Zero‑Budget Release Notes: A Case Study in Free‑Tier LLM Automation

Description and use

A zero‑budget release‑note generator using a free LLM and server, running unattended on schedule.

Problem

Weekly library releases required a human to read 40–50 technical commit messages and rewrite them into user‑friendly notes, incurring time and risk of errors.

Outcome

The bot produced concise, categorized release notes without any paid API usage, running unattended on a free server and handling rate‑limit failures gracefully.

Workflow

• Collect commit subjects between two git tags via git log --oneline
• Chunk commits into 25‑item groups
• Send each chunk to an OpenAI‑compatible chat endpoint with a system prompt that returns a JSON array of categorized bullets
• Validate and aggregate the JSON responses
• Render the aggregated items into a Markdown file
• Schedule the script with cron for unattended runs

Evidence

The project used MonkeyCode's free model access and its free server option for hosting.

Source

The implementation is a single Python file with three responsibilities: collect commits, summarize chunks, and write the notes file.

Source