Skip to content
This repository was archived by the owner on Jan 30, 2026. It is now read-only.

Latest commit

 

History

History
178 lines (120 loc) · 3.3 KB

TEMPLATE.md

File metadata and controls

178 lines (120 loc) · 3.3 KB

AI Coding Agent Workflow Template

TL;DR Cheat Sheet

  • Best for:
  • Avoid when:
  • Key prompt:
  • Biggest win:
  • Biggest risk:

Use this template to share practical, repeatable AI coding agent workflows. The goal is to help others understand when to use an agent, how to set it up, and what good looks like when it's working well.

1. Workflow Overview

Workflow Name A short, descriptive name that makes it obvious what this does.

Primary Goal What problem does this workflow solve? What outcome does it reliably produce?

Audience

  • Backend
  • Frontend
  • Mobile
  • Infra / Platform
  • Data / ML
  • Full-stack
  • Other: __________

When to Use This Signals that this workflow is a good fit (e.g., repetitive changes, large surface area, unfamiliar codebase, etc.).

When NOT to Use This Clear anti-patterns or cases where manual work is safer/faster.

2. Prerequisites & Setup

Codebase Assumptions

  • Repo size / structure expectations
  • Languages / frameworks
  • Test coverage expectations

Tools / Agents Used

  • Agent / IDE (e.g., Cursor, Copilot, Claude Code, custom agent)
  • Supporting tools (linters, formatters, CI checks, etc.)

Required Context Provided to the Agent What you explicitly load or paste before starting.

  • Key directories or files
  • Architecture docs or READMEs
  • Coding standards or style guides

3. Agent Role & Framing

Agent Role Definition How you frame the agent's responsibility. Example: "You are a senior backend engineer focused on safe, incremental refactors."

Constraints Given to the Agent

  • What it must not change
  • Required patterns or APIs
  • Performance / security constraints

Success Criteria How you judge whether the agent did a good job.

4. Step-by-Step Workflow

Be concrete and sequential. Someone else should be able to follow this exactly.

Step 1 – Initial Prompt

<Initial prompt goes here>

What happens: Describe what the agent should do in response.

What to check: What you verify before proceeding.

Step 2 – [Next Action]

<Follow-up prompt or action>

What happens: What to check:

Continue for each step in the workflow...

5. Quality Gates & Validation

Before Accepting Agent Output:

  • Checkpoint 1
  • Checkpoint 2
  • Checkpoint 3

Testing Strategy How you verify the changes work.

Rollback Plan What you do if something goes wrong.

6. Results & Learning

Typical Outcomes What success looks like (with numbers/metrics if available).

Common Pitfalls What typically goes wrong and how to avoid it.

Iteration Tips How to improve the prompt or workflow based on experience.

7. Example Run

Context: Brief description of the codebase/problem.

Initial State:

<Code or description before>

Agent Prompts Used:

Prompt 1: ...
Prompt 2: ...

Final State:

<Code or description after>

Outcome: What changed, what worked, what didn't.

8. Related Workflows

  • [Link to related workflow 1]
  • [Link to related workflow 2]

9. Contributors & Changelog

  • Author: [Your Name]
  • Date: YYYY-MM-DD
  • Version: 1.0

Changelog:

  • v1.0 - Initial workflow documented