AgentSkillsCN

implement_plan

借助Gemini CLI工具,分阶段执行既定的技术方案,从想法/计划出发,通过文件交互、命令执行与任务管理,实现方案落地。

SKILL.md
--- frontmatter
name: implement_plan
description: Implement an approved technical plan from thoughts/plans/ with phase-by-phase execution, leveraging Gemini CLI tools for file interaction, command execution, and task management.
disable-model-invocation: true
allowed-tools: read_file, write_file, run_shell_command, search_file_content, glob, write_todos, delegate_to_agent
argument-hint: "[plan-file-path]"

Implement Plan

You are tasked with implementing an approved technical plan from thoughts/plans/. These plans contain phases with specific changes and success criteria.

Directory Structure:

  • thoughts/tickets/ - Original feature requests and task descriptions
  • thoughts/plans/ - Implementation plans (the files you'll be executing)
  • thoughts/research/ - Supporting research and investigation notes

Getting Started

When given a plan path:

  • Read the plan completely and check for any existing checkmarks (- [x])
  • Read the original ticket and all files mentioned in the plan
  • Read files fully - never use limit/offset parameters
  • Use write_todos to track your progress
  • Start implementing if you understand what needs to be done

If no plan path provided, ask for one.

Implementation Philosophy

Plans are carefully designed, but reality can be messy. Your job is to:

  • Follow the plan's intent while adapting to what you find
  • Implement each phase fully before moving to the next
  • Verify your work makes sense in the broader codebase context
  • Update checkboxes in the plan as you complete sections

If you encounter a mismatch:

  • STOP and present the issue clearly:
    code
    Issue in Phase [N]:
    Expected: [what the plan says]
    Found: [actual situation]
    Why this matters: [explanation]
    
    How should I proceed?
    

Verification Approach

After implementing a phase:

1. Run Success Criteria Checks

Use technology-appropriate commands:

Node.js/JavaScript: npm test, npm run lint, npm run build Python: pytest, black --check ., mypy . Go: go test ./..., golangci-lint run, go build Rust: cargo test, cargo clippy, cargo build Make-based: make test, make lint, make build

2. Fix Issues and Update Progress

  • Address any failures before moving to the next phase
  • Update checkboxes in the plan file using the Edit tool
  • Update your TodoWrite list

If You Get Stuck

  1. Investigate First - Read all relevant code completely
  2. Delegate to research agents (e.g., codebase_investigator, codebase_locator, codebase_pattern_finder) for targeted help
  3. Present Issues Clearly - Don't guess, ask for clarification

Resuming Work

If the plan has existing checkmarks:

  • Trust that completed work is done correctly
  • Pick up from the first unchecked item
  • Verify previous work only if something seems off

Key Principles

  1. Deep Understanding Before Action - Read files completely
  2. Follow Intent, Not Just Instructions - Adapt to reality
  3. Maintain Quality - Run verification commands
  4. Communicate Clearly - Update checkboxes and todos
  5. Keep Momentum - Don't get stuck on minor details