Github 5 bullet points

This repo creates a weekly sum up from your Github activities, in 5 bullet points.

Why? Please search on Google for Send 5 bullets of what you accomplished last week

Important

Security concern? For now, this can only read your PUBLIC interaction, meaning only public repos are taken into account.

Demo:

How it works

• We user Github API /users/{username}/events to list all your public events (no private events for now)
• Convert it to prompt
• Ask LLM to summarize, you can use any OpenAI-compatible API provider
• Push the summarization to README.md of your https://github.com/{your_username}/{your_username}

How to install it?

1. Make sure you created a repo at https://github.com/{your_username}/{your_username}. It can be either a blank repo or with an existing README
2. Fork this repo
3. Enable Github Action on your forked repo
4. Go to https://github.com/settings/personal-access-tokens and create a new PAT (personal access token)

• Token name: 5-bullet-points (or anything else you want)
• Resource owner: you (your personal account)
• Expiration: Recommended to set it to No expiration, or if you just want to test it, set to 30 days
• Under Repository access: Only select repositories, then select the repo in step 1
• Permissions --> Contents --> Access: Read and Write
• Click on Generate token
• Keep the generated token somewhere safe, will need to use it in later steps

5. Go back to the forked repo, go to Settings --> Secrets and variables --> Actions
6. Create a new Repository secrets:

• Name: GH_PAT_WRITE
• Value: The token from step 4

7. Depending on which LLM inference provider you want to use, follow to guide accordingly:

• Recommended: use DeepSeek-R1-Qwen-14B Hugging Face Inference Provider, create 3 more secrets
  • OAI_COMPAT_URL: https://router.huggingface.co/{provider}
    Example: https://router.huggingface.co/novita
  • OAI_COMPAT_TOKEN: The token created from https://huggingface.co/settings/tokens (remember to check Make calls to inference providers when create it)
  • OAI_COMPAT_EXTRA_BODY: {"model":"deepseek/deepseek-r1-distill-qwen-14b"} (or other models on HF hub)
• If you wants to use OpenAI:
  • OAI_COMPAT_URL: https://api.openai.com/v1
  • OAI_COMPAT_TOKEN: Your OpenAI key
  • OAI_COMPAT_EXTRA_BODY: {"model":"gpt-4o"} (or other models)
• Same for other providers

Optionally: Create a new Repository variables (not a secret) to customize the output README file:

• Name: README_TEMPLATE
• Value: See example in config.ts

Then finally, go to Actions tab in the forked repo --> "Run it" --> "Run workflow" on the right side

Cronjob

A default cronjob is setup for every Friday at 7pm (GMT+0). You can modify it in .github/workflows/run-it.yml