🎁 Get the FREE AI Skills Starter Guide β€” Subscribe β†’
BytesAgainBytesAgain
πŸ¦€ ClawHub

feature-modification-workflow

by @liancheng-zcy

Structured workflow to analyze, plan, and execute feature modifications by scenario complexity, ensuring controlled, incremental, and risk-aware changes.

Versionv1.0.0
Downloads557
TERMINAL
clawhub install feature-modification-workflow

πŸ“– About This Skill


name: feature-modification-workflow description: A structured workflow for modifying existing features. TRIGGER when: user says "I want to change a feature", "optimize xxx", "check if xxx has room for improvement", "fix a bug", or any request involving feature modifications. Choose the appropriate workflow based on scenario complexity: simple issues go straight to code, complex issues follow the full process.

Feature Modification Workflow

A standardized workflow for modifying existing features, ensuring changes are well-founded, controlled, and incremental.

Core Principles

Understand First, Then Modify β€” Before any feature change, understand the relevant code to ensure the change direction is correct.

Choose Workflow by Scenario β€” Simple issues get simplified workflow, complex issues get full workflow. No unnecessary overhead.

Check Docs When Needed β€” Only check latest documentation when dealing with complex technical details or specific API usage. Simple changes don't need this.

Conservative Optimization β€” Optimize without changing existing flows and core logic. Avoid high-risk large-scale refactoring.

Confirm Before Execute β€” Complex changes require user confirmation before development begins.


Step 1: Determine Scenario

After receiving a user request, first determine which scenario it falls into, then choose the corresponding workflow:

| Scenario | Characteristics | Workflow | |----------|-----------------|----------| | Simple Issue | Change color, text, spacing, config values | Simplified: Read code β†’ Change directly | | Bug Fix | Fix errors, exceptions, unexpected behavior | Must analyze cause. Simple bugs use simplified workflow, complex bugs use full workflow | | Complex Issue | New features, logic optimization, multi-file changes, architecture involved | Full 5-step workflow | | Refactoring | Core flow changes, large-scale code adjustments | Full 5-step workflow |


Scenario 1: Simple Issue (Simplified Workflow)

What Counts as Simple Issue

  • UI adjustments: change color, font size, spacing, borders
  • Text modifications: change copy, hints, labels
  • Config modifications: change defaults, parameters
  • Single-file small changes: doesn't affect other modules
  • Simplified Workflow

    1. Read relevant code (only the part to change)
    2. Make the change
    3. Inform user it's done
    

    Example

    User: Change the login button color to blue

    AI:

  • Read button code β†’ Find color setting β†’ Change to blue β†’ Done

  • Scenario 2: Bug Fix (Must Analyze Cause)

    Core Requirement

    Bug fixes must first locate the cause. Don't change code based on guessing.

    Simple Bug (Simplified Workflow)

    Characteristics: Cause is clear, small change scope, doesn't affect other features

    1. Analyze current state: locate bug cause
    2. Fix directly
    3. Inform user it's done
    

    Complex Bug (Full Workflow)

    Characteristics: Cause unclear, involves multiple modules, may affect other features, requires architecture changes

    Follow the full 5-step workflow
    

    Judgment Criteria

    | Simple Bug | Complex Bug | |------------|-------------| | Clear error message, can locate directly | Cause unclear, needs investigation | | Small change scope, 1-2 places | Involves multiple files/modules | | Doesn't affect other features | May cause chain reactions | | Doesn't involve core logic | Needs to change core flow |


    Scenario 3: Complex Issue (Full 5-Step Workflow)

    What Counts as Complex Issue

  • Adding new features
  • Optimizing logic/performance
  • Changes involving multiple files
  • Architecture adjustments
  • Requires design proposals
  • Full Workflow

    #### Step 1: Understand the Project Globally

    After receiving a feature change request, don't start coding directly. First do the following:

    1. Understand project structure - Use Glob to view directory structure - Identify core modules and dependencies - Find relevant config files, entry files

    2. Read key files - Project documentation (CLAUDE.md, README.md, etc.) - Core module implementation files - Files directly related to the change target

    3. Understand existing flows - How data flows - How modules communicate - What the user interaction flow is

    Output Format:

    ## Project Overview
    
  • Project type: xxx
  • Core modules: xxx, xxx, xxx
  • Data flow: xxx β†’ xxx β†’ xxx
  • Related Modules

  • Module A: responsible for xxx
  • Module B: responsible for xxx
  • Dependencies: A β†’ B β†’ C
  • #### Step 2: Analyze Current State

    After understanding the project, analyze the current state of the change target:

    1. Current implementation analysis - How is current code implemented - What technologies/patterns are used - What key design decisions were made

    2. Identify strengths - What's good about current implementation - Which designs are reasonable and worth keeping

    3. Identify bottlenecks - Where are performance bottlenecks - Where are UX bottlenecks - Where are maintainability bottlenecks

    4. Identify weaknesses - Code quality issues - Design flaws - Extensibility limitations

    Output Format:

    ## Current State Analysis

    Current Implementation

    [Describe current implementation]

    Strengths

  • xxx
  • xxx
  • Bottlenecks/Weaknesses

  • xxx: [specific description]
  • xxx: [specific description]
  • #### Step 3: Propose Optimization Plan

    Based on analysis, propose optimization plan. Key constraint: don't change existing flow

  • Optimization should be incremental, not a rewrite
  • Keep core logic and data flow
  • Only make local improvements
  • Output Format:

    ## Optimization Plan

    Overview

    [Briefly describe optimization approach]

    Optimization Points

    1. xxx: [specific change, expected effect] 2. xxx: [specific change, expected effect]

    What Not to Change

  • xxx: [why to keep]
  • Risk Assessment

  • Change scope: [large/medium/small]
  • Affected modules: xxx, xxx
  • Note: xxx
  • #### Step 4: List Plan and Ask User

    Break down the optimization plan into specific steps, list the plan and ask user for confirmation:

    Output Format:

    ## Development Plan

    Step List

    1. [ ] Step 1: xxx 2. [ ] Step 2: xxx 3. [ ] Step 3: xxx

    Confirmation Questions

  • Does the above plan meet your expectations?
  • Any steps need adjustment or addition?
  • Can I start execution?
  • Wait for user response. Don't start development on your own.

    #### Step 5: Execute Step by Step After Confirmation

    After user confirms:

    1. Use TodoWrite to create task list, track progress 2. Develop step by step 3. Mark completion and inform user after each step 4. Report issues promptly, don't bypass problems

    During development:

  • Keep communication, ask user for important decisions
  • Complete one step before starting next
  • If plan needs adjustment, ask user first

  • Scenario 4: Refactoring (Full 5-Step Workflow)

    Refactoring must follow the full workflow because of high risk and large impact scope.

    Characteristics of refactoring:

  • Changes to core flow
  • Large-scale code adjustments
  • Involves multiple modules
  • May affect existing features

  • Conditional Step: Check Documentation

    Not all changes need documentation check. Only check in these cases:

    When to Check Documentation

    1. Complex technical details - Audio processing algorithms (sample rate conversion, silence detection) - Network protocol details (WebSocket frame format, Protobuf encoding) - Multi-threading/concurrency issues (thread safety, lock mechanisms)

    2. Unsure about specific API usage - Framework API parameter meaning unclear - Library function best practices - API changes after version updates

    3. Involves specifications or standards - Data format specifications - Protocol standard documents - Security/compliance requirements

    When NOT to Check Documentation

  • UI layout adjustments (spacing, colors, fonts)
  • Simple logic changes (conditionals, parameter adjustments)
  • Already familiar API usage
  • Similar implementation exists in project for reference
  • Documentation Check Tools Priority

    1. Context7 (mcp__context7__resolve-library-id + mcp__context7__query-docs) - Latest docs for programming libraries/frameworks - PySide6, React, numpy, etc.

    2. WebSearch - Protocol specifications, technical articles - WebSocket protocol, best practice articles


    Quick Reference

    Scenario-Workflow Mapping

    | Scenario | Workflow | |----------|----------| | Simple Issue | Read code β†’ Change directly | | Simple Bug | Analyze cause β†’ Fix directly | | Complex Bug | Full 5 steps | | Complex Issue | Full 5 steps | | Refactoring | Full 5 steps |

    Full 5 Steps

    | Step | Action | Output | |------|--------|--------| | 1 | Understand globally | Project overview | | 2 | Analyze current state | Strengths/bottlenecks/weaknesses | | 3 | Propose plan | Optimization points/what not to change | | 4 | List plan and ask | Step list + confirmation questions | | 5 | Execute after confirmation | Develop step by step |

    Conditional: Check documentation (for complex technical issues)


    Remember: Determine scenario first, then choose workflow. Handle simple issues simply, plan complex issues carefully.

    πŸ’‘ Examples

    User: Change the login button color to blue

    AI:

  • Read button code β†’ Find color setting β†’ Change to blue β†’ Done