💻✨ AI for Coding
Short, actionable tips to help developers get more value from AI coding tools. Originally shared as an internal Slack series.
Tip #1: Give Copilot a memory
Set up an instruction file so the agent knows your conventions from the start.
Tip #2: Every AI mistake is a rule
When Copilot gets something wrong, add it to your instructions instead of just fixing it.
Tip #3: Use Copilot as a sparring partner
Think together before writing code — use plan mode to explore approach and tradeoffs.
Tip #4: Paste the error, get the explanation
Don’t Google cryptic errors. Paste them into chat and ask for the explanation first.
Tip #5: What actually belongs in your instruction file?
Less is more: hand-written, non-obvious rules beat auto-generated descriptions.
Tip #6: Your terminal just got an AI agent
Copilot CLI brings agentic coding to your terminal — reads files, runs commands, makes changes.
Tip #7: Let Copilot review every PR — automatically
Set up branch rulesets so Copilot reviews pull requests without anyone asking.
Tip #8: Teach your AI agent repeatable skills
Package step-by-step procedures as SKILL.md files the agent loads on demand.
Tip #9: Where to put your skills (and where they'll actually be found)
Project skills vs. personal skills, and how the agent discovers them.
Tip #10: Why your skills might not be triggering
The agent often ignores skills unless you tell it to look for them first.
Tip #11: What your agent already knows how to do
Built-in tools for file editing, search, shell commands, and GitHub — no setup needed.
Tip #12: MCP — the USB-C of AI tools
One open standard for connecting agents to external systems across all tools.
Tip #13: Give your agent a browser
Chrome DevTools MCP lets the agent click, inspect, and test web UIs from your terminal.
Tip #14: Your CLI tools are already AI tools
gh, git, jq, curl work without MCP — the agent just calls –help.
Tip #15: MCP servers eat your context window
Every connected server costs tokens every turn. Be selective.
Tip #16: Explore before you edit
Use the agent as a research assistant to understand unfamiliar code before changing anything.
Tip #17: Keep sessions short and focused
One task per session. Context fills up — fresh starts beat accumulated corrections.
Tip #18: Commit before you let the agent loose
Git stash or commit first, so the agent’s changes are always one git diff away from a clean state.
Tip #19: Get a second opinion from a fresh agent
Open a second session to re-ask the same question without baggage from earlier attempts.
Tip #20: Give your agents their own workspaces
Use isolated workspaces so multiple agents can work on the same repo without stepping on each other.
Tip #21: Meet OpenCode — the open source coding agent
A provider-agnostic coding agent that works with your existing Copilot license and 75+ other providers.
Tip #22: Use scripts for repeatable tasks
Deterministic tasks deserve deterministic scripts — not AI improvisation every session.
Tip #23: Write scripts that guide the agent
Script output goes straight into the agent’s context window. Design it to say what happened and what to do next — not just what ran.
Tip #24: Automate the boring parts with hooks
Hooks are shell commands that fire automatically at agent lifecycle points — before/after a tool runs, on session start, on finish. Unlike instruction files, they’re deterministic. Enforce your venv rule, run tests after edits, get a desktop notification when the agent is done.
Tip #25: Your engineering skills didn't disappear — they shifted
Less syntax, more architecture. Less typing, more judgment. AI rewards good engineering — but only if you stay sharp.
Tip #26: Before you wire AI agents to CI/CD, read this
The Clinejection attack showed how a malicious PR description can hijack an AI agent with CI/CD access. Know the risk before you automate.