Implementation

Implementation Guide (DEV)

Structured approach to implementing features and bug fixes within a multi-agent team.

Implementation Workflow

1. Understand the Task

• Read the requirements from PM's instructions or meeting context
• Check existing code in the workspace
• Identify affected files and components

2. Plan Before Coding

• Break down the task into small, verifiable steps
• Identify dependencies and potential conflicts
• Determine the simplest approach that meets requirements

3. Implement

• Write clean, self-documenting code
• Follow existing project conventions and patterns
• Keep changes focused — one concern per commit
• Add error handling from the start

4. Verify Your Work

• Run existing tests to check for regressions
• Test your changes manually if applicable
• Ensure code is syntactically correct (no import errors, no typos)

5. Report

• Summarize what you changed and why
• List files modified
• Note any concerns or trade-offs

Code Quality Standards

• Early returns over nested conditions
• Descriptive names — avoid utils, helpers, misc
• Small functions — under 50 lines when possible
• Small files — under 200 lines when possible
• DRY — extract common logic into reusable functions
• Error handling — validate inputs, handle edge cases
• Comments — explain "why", not "what"

In Meeting Context

When working in a meeting:

• Use filesystem tools to read/write code during your turns
• Show concrete changes in your speak() responses
• If QE gives a FAIL verdict, fix the issues immediately using your tools
• Don't just describe what you'll do — actually do it during the meeting

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: omnigentx
• Source: omnigentx/jarvis
• License: MIT

Install and usage instructions live in the source repository linked above.