🎤 This mode is just for fun lah. Some technical nuances might be simplified or lost in translation. We also used AI to help write these (where got time to do one by one). For the accurate stuff, switch back to the other modes.
What is GitHub?
GitHub is a free online platform where people store and collaborate on projects. Think of it as a supercharged Google Drive - except it also tracks every change anyone makes, lets you create task lists, discuss ideas, and automate workflows.
GitHub is a cloud-hosted Git platform with integrated project management tooling: Issues for task tracking, Projects for kanban/table views, Actions for CI/CD automation, and Discussions for async communication. Free for unlimited public and private repos.
Eh bro, you know GitHub or not? It's like your Google Drive on steroids lah. Got track changes, got task list, got everything. Your whole team can see who do what, when they do, how they do. Best part? Free one. Like the best hawker centre of project management - unlimited servings, no need to pay extra.
Repositories
A repository (or "repo") is a project folder. It holds all your files and tracks every change made to them. You can have as many repos as you like - one per project.
A Git repository stores your project's source tree plus full history. GitHub adds Issues, Projects, Actions, and a web UI on top of the core VCS functionality.
Repository ah, basically just your project folder lor. But this one power - it remembers every single change you ever make. Like your mother remembers every mistake you made since primary school. Cannot delete history one, so don't anyhow commit embarrassing stuff ah.
Issues
Issues are like sticky notes or to-do items. You create one for each task, bug, or idea. You can add descriptions, assign them to people, tag them with labels (like "urgent" or "design"), and track their progress.
Issues are GitHub's core work-tracking primitive. They support markdown descriptions, labels, assignees, milestones, and linked pull requests. Each issue gets a unique number and its own comment thread.
Issues are like your sticky notes on the fridge lah, but digital. "Buy groceries", "Fix the aircon", "Design poster for event". Can tag people some more - "Eh Ahmad you do this one". Got labels like "urgent" or "shiok already done". Very the organised sia.
Projects (Boards)
Projects give you a visual board - like Trello or a wall of sticky notes. You can drag tasks between columns like "To Do", "In Progress", and "Done". You can also switch to a table view, a timeline, or a roadmap.
GitHub Projects (v2) provides kanban, table, timeline, and roadmap views over issues and PRs. Supports custom fields, grouping, filtering, saved views, and automation rules (e.g. auto-move on close).
This one is basically your Trello board lah. Got columns - "Haven't Start", "Doing", "Done Already". Drag and drop your tasks around. Very satisfying when you move something to "Done" - like that feeling when you clear your plate at buffet. Shiok.
Discussions
Discussions are like a forum or group chat for your project. They're the right place for open-ended questions, announcements, or brainstorming - things that aren't specific tasks yet.
GitHub Discussions is a built-in forum for async communication. Categories, threaded replies, polls, and answered-question workflows. Keeps conversational content separate from actionable Issues.
Discussions is like your group chat, but got structure one. Not like WhatsApp where everything gets buried under 200 messages of "noted" and sticker spam. Can separate into topics properly - like the uncle at kopitiam who actually stays on topic. Rare but useful.
What are Coding Agents?
Coding agents are AI assistants that live inside your code editor. You type what you want in plain English - "create a task list for the event", "write a project overview", "organise these notes" - and the AI does it. They're not just for programmers; they work with any text-based task.
Coding agents are LLM-powered tools integrated into IDEs (VS Code, Antigravity, Cursor) with capabilities beyond autocomplete: multi-file editing, terminal access, tool-use, and agentic workflows. They can create issues, draft documents, refactor code, and orchestrate multi-step tasks based on natural language prompts.
Coding agents ah, you know those robot helpers inside the code editor? You just type what you want in plain English - "create task list for Deepavali event", "write the meeting minutes" - then the AI go and do for you. Steady pom pi pi. It's like having an intern that never complain and works 24/7. Dream come true sia.
GitHub Copilot
GitHub's own AI assistant. It suggests code as you type and can answer questions about your project. Available free for students, open-source contributors, and with limited usage for everyone.
GitHub Copilot offers inline completions, chat, and an agent mode. Integrates with Issues, PRs, and the GitHub API. Copilot Workspace can plan and implement multi-step changes from an issue description.
GitHub's own AI helper. This one built-in one - like the complimentary kacang at the bar. Free for students and open-source people. It can suggest code, answer questions, even plan whole projects. Not bad for something you don't need to belanja.
Claude (in Antigravity)
An AI assistant made by Anthropic. In Antigravity (Google's free code editor), Claude can help you with writing, planning, and creating content - not just code. Describe what you need and it does the work.
The latest Claude 4.6 models are available in Antigravity's agent mode. Excels at reasoning, planning, and multi-step tasks. Can create files, run terminal commands, and iteratively refine work based on feedback.
Claude is made by Anthropic - basically the atas AI lah. Very good at thinking and planning. In Antigravity (Google's free editor), you just tell Claude what you want, it go and whack for you. Like having that one friend who always "can can, I settle" - and actually settles.
Agent Modes
Most AI editors have three modes: Ask (just chat), Edit (make specific changes), and Agent (do multiple things autonomously). Agent mode is the most powerful - it can plan, create files, and complete multi-step tasks on its own.
Modern AI editors typically offer: Ask (chat, no edits), Edit (targeted inline changes), and Agent (autonomous multi-file operations with tool use). Agent mode is the most effective for project management tasks - it can create multiple files, run commands, and iterate.
Got three modes ah: Ask (just teh tarik chat only), Edit (make small small changes), and Agent (this one the siao on mode - it plan, it create, it do everything by itself). Agent mode is like engaging contractor - you say what you want, they go and build. You just sit there and supervise lor.
Project Management with GitHub
Here's how to set up and use GitHub as your project management hub. If you're familiar with Trello, Asana, or Monday.com - this works similarly, but it's free and deeply integrated with your actual work.
GitHub Projects provide a comparable feature set to Jira, Linear, or Asana - with the advantage of being natively integrated with your codebase, CI/CD, and collaboration tools.
OK now the real deal lah - how to use GitHub for project management. If you ever used Trello or Asana before, same same but different. This one free and connected to everything. Like getting the economy rice uncle to cater your whole event - cheap, good, and reliable.
Create a Project Board
Go to your GitHub profile → "Projects" tab → "New project". Choose the "Board" layout. You'll get columns like "To Do", "In Progress", and "Done" - just like a physical task board.
Navigate to your org or profile → Projects → New project. Choose Board (kanban), Table, or Roadmap layout. You can switch between views at any time. Add custom fields (Status, Priority, Sprint, etc.) from the project settings.
First thing first - go your GitHub profile, click "Projects", then "New project". Choose "Board" layout. Wah, confirm got the nice columns like "To Do", "In Progress", "Done". Feel like you're in some atas corporate office already, but actually you're in shorts at home.
Create Issues for Tasks
For each task or to-do, create an Issue. Give it a clear title like "Design the event poster" or "Book the venue". Add a description with details. You can use checklists inside an issue to break it into smaller steps.
Create Issues in your repo for each work item. Use markdown for descriptions - checklists (- [ ]), headings, links. Reference other issues with #123. Use issue templates for consistency across recurring task types.
For every task, make one Issue. Title must clear one ah - "Design the poster", not "Do that thing". Put description with all the details. Can even put checklist inside - like your mother's to-do list but digital. Nobody can pretend they didn't see it.
Organise with Labels & Milestones
Labels are coloured tags you put on issues - like "urgent", "design", "content", or "bug". Milestones group issues into goals with a due date, like "Website Launch" or "Q2 Event Series". Together, they help you filter and prioritise.
Define a label taxonomy (type: bug/feature/chore, priority: P0-P3, area: frontend/backend/design). Milestones map to releases or sprints with due dates. Use label-based filters in your Project views for focused workflows.
Labels are like those colour stickers you put on your files in school lah. "Urgent" in red, "Design" in blue, whatever you want. Milestones is like deadline grouping - "Hari Raya Event" or "Q2 Stuff". Very kiasu move but it works.
Use Automations
GitHub can automatically move tasks for you. When someone marks an issue as done, it moves to the "Done" column. When a new issue is created, it appears in "To Do". You set this up once and it just works.
GitHub Projects supports built-in automations: auto-add new issues, auto-set status when issues close, auto-archive completed items. For more complex workflows, use GitHub Actions with the projects API.
This one best part - set it once, then autopilot. Close an issue? Auto move to "Done". Create new issue? Auto appear in "To Do". It's like having that one team member who actually follows through without you having to chase. Walao, if only real people were like this.
Assign & Collaborate
Assign issues to team members, leave comments to discuss them, and @mention people to get their attention. Everyone can see what's happening, what's done, and what needs attention - all in one place.
Use assignees for ownership, @mentions for notifications, and issue comments for async discussion. Pin important issues, use reactions for lightweight feedback, and convert comments to new issues when scope creeps.
Assign task to people, @mention them so they kena notified. Leave comments to discuss. It's like a WhatsApp group but actually productive - no random memes, no "noted with thanks", just actual work getting done. Revolutionary concept sia.
Suggested Agent Prompts for PM
These are prompts you can give to your AI coding agent (in Antigravity, VS Code, or similar) to help with project management tasks. Copy, adapt, and use them as starting points.
"Break down [project/feature] into a list of GitHub Issues. For each issue, include a clear title, description, acceptance criteria, and suggested labels."
"Create a milestone called [name] with a due date of [date]. Suggest which of these issues should be included and why."
"Write a project README that explains [project] for new team members. Include the purpose, how to get started, key links, and who to contact."
"Create a GitHub Issue template for bug reports. Include sections for: description, steps to reproduce, expected vs actual behaviour, screenshots, and environment info."
"Summarise the current state of this project: how many issues are open, in progress, and closed. Highlight any blockers or overdue items."
"Review these open issues and suggest a priority ranking. Consider impact, effort, and dependencies. Flag any that are blocking others."
"Based on the issues closed this sprint, draft a retrospective summary: what went well, what was challenging, and what to improve next time."
"Create an onboarding checklist as a GitHub Issue for a new team member joining [project]. Include setup steps, key docs to read, and people to meet."
Beyond Code - PM for Any Project
GitHub isn't just for programmers. If your work involves tasks, deadlines, and collaboration, it can be your project management hub too - and it's entirely free.
GitHub's PM tooling is domain-agnostic. Issues and Projects work for any workflow that benefits from structured task tracking, async discussion, and automation - not just software development.
Eh don't think GitHub only for coders ah. If your work got tasks, got deadlines, got people who need to collaborate - this works for you also. And it's free leh! Stop paying for all those fancy project management tools when you can get the same thing without burning a hole in your pocket.
Event Planning
Create issues for each task: book venue, confirm speakers, design flyers, arrange catering. Use milestones for key dates and a board to track progress across the team.
Content Calendars
One issue per piece of content - blog posts, social media, newsletters. Label by platform and status. Use the timeline view to see your publishing schedule at a glance.
Research Projects
Track literature reviews, experiments, and findings as issues. Use checklists for methodology steps. Discussions for brainstorming. Everything is versioned and searchable.
Client Work & Freelancing
One repo per client. Issues for deliverables. Milestones for project phases. The full history shows exactly what was done and when - useful for billing and accountability.
Marketing Campaigns
Plan campaign tasks, creative assets, and launch activities. Use labels for channels (email, social, print) and automations to move tasks as they progress.
Team Coordination
Use a shared project board for cross-team visibility. Each team member sees their assigned tasks. Discussions replace endless email chains. Everything lives in one place.
Limitations to Be Aware Of
Learning curve
GitHub's interface can feel overwhelming at first - it was designed for developers. Give yourself a couple of sessions to explore. The AI agent can help you navigate and set things up.
AI isn't always right
AI agents can occasionally misunderstand your intent or produce inaccurate information. Always review what they generate - especially issue descriptions, priorities, and status summaries.
Privacy considerations
If your project contains sensitive information, use a private repository. Be mindful that AI agents may process your content through external APIs - check your organisation's data policies.
Not a complete replacement
GitHub PM tools are powerful but don't replace human judgement. Use them to organise and accelerate your work - not to make decisions about priorities, people, or strategy.
Related Resources
Set up the full Astro + Cloudflare stack to build and deploy websites for free - then use these PM techniques to manage the project.
Agentic Website Guide →