Skip to content

Kavya-24/Helpful-LLM-Routines

Repository files navigation

Helpful LLM Routines

A collection of reusable LLM routines for common engineering workflows. Each routine is a YAML file that defines a task / routine / that LLM (Claude, ChatGPT) can run on demand or on schedule — no code required.

What is a scheduled routine?

A scheduled routine is a structured prompt with named inputs. You fill in the inputs for your environment, hand it to Claude, ChatGPT or any other LLM executor connected with MCPs and let it execute in background at your decided frequency. Routines are designed to be:

  • Generic — no hardcoded usernames, emails, or org names
  • Safe — nothing is sent or posted without an explicit "send" step
  • Composable — inputs can be shared across routines

Routines

File Connectors What it does
engineering_slack_daily_brief.yaml Slack, GitHub, Notion Scans Slack for mentions from the last day and prepares draft replies by category (PR review, doc review, Q&A)
engineering_standup_generator.yaml GitHub, Slack Generates a Yesterday / Today / Blockers standup from your GitHub activity and Slack messages

How to use

Option A — /run-routine skill (recommended)

If you're in Claude Code, run:

/run-routine engineering_slack_daily_brief.yaml

Claude will read the routine, infer what it can, ask you to confirm or fill in the remaining inputs in one shot, then show you the final prompt and ask before executing.

Option B — manual

  1. Open a routine YAML and fill in the routine with your identifiers
  2. Copy the prompt and set up the routine in Claude Code Routines / ChatGPT Schedules

Schema

name: Human-readable name
tags:
  - engineering
description: >
  One or two sentences describing what the routine does.
connectors:
  - ConnectorName        # integrations the LLM needs access to
inputs:
  - name: input_name
    description: What this value is and where to find it.
routine: |
  The prompt LLM executes. Reference inputs as {{input_name}}.

Connectors

Each routine declares the integrations it needs under connectors. Make sure LLM has access to those connectors before running.

Connector Used for
Slack Reading messages and threads, posting drafts
GitHub Reading PRs, issues, CI status, commit history
Notion Reading linked documents and pages

Contributing

  1. Copy an existing YAML as a starting point.
  2. Follow the schema above.
  3. No PII in the file — use {{input_name}} placeholders in the routine block.
  4. Open a pull request with a short description of the use case.

License

MIT

Releases

No releases published

Packages

 
 
 

Contributors