mongodb / chatbot

MongoDB Chatbot Framework. Powered by MongoDB and Atlas Vector Search.
https://mongodb.github.io/chatbot/
Apache License 2.0
106 stars 48 forks source link

(EAI-348) Summarize Release Artifacts #416

Closed nlarew closed 1 month ago

nlarew commented 1 month ago

Jira

Changes

How to use

Env Vars

You need the following new .env vars + the rest of .env.example

The GitHub token you can make yourself (https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens). I can send you jira creds.

First time CLI setup

cd mongodb-artifact-generator
npm install
npm link # makes mongodb-ai binary available in your shell

Generate release note artifact summaries

mongodb-ai generateReleaseNotes \
  --runId="my-run \
  --releaseInfo="./context/release-notes/atlas-cli-releaseInfo.yaml" 

The POC run takes around 45 seconds to summarize 31 artifacts. There are console logs to track progress.

When the run is complete, the generated artifacts are available in the runlogs/generateReleaseNotes/<runId> directory.

nlarew commented 1 month ago

i wonder if we could add a LLM function call that tries to determine if a change is public facing. think chain-of-thought prompting could figure that out reasonably well.

maybe also a function call that tries to determine type of change. i/e bug fix, feature, etc.

Good idea! I'll look at implementing that in the changelog piece I'm working on now