AgentSkillsCN

plan-implementation

为某项功能或变更制定结构化的实施计划。分析目标,识别需修改的文件,检查代码模式,评估潜在风险。在完成调研后、开始编码前使用此功能。

SKILL.md
--- frontmatter
name: plan-implementation
description: Create a structured implementation plan for a feature or change. Analyzes goal, identifies files to modify, checks patterns, identifies risks. Use after researching, before coding.

Plan Implementation

Creates an actionable checklist-style plan from research and context.

Inputs

InputTypeDefaultPurpose
goalstringrequiredWhat to implement (e.g., "Add Redis caching to billing API")
projectstringautoProject to plan for
issue_keystring-Jira issue key if for specific ticket
constraintsstring-Requirements (e.g., "must be backwards compatible")

Workflow

1. Load Developer Persona

  • persona_load("developer") — for Jira tools

2. Detect Project

  • Infer from cwd; default automation-analytics-backend

3. Get Jira Issue (if issue_key)

  • jira_view_issue(issue_key=issue_key) — summary, acceptance criteria

4. Search Related Code

  • code_search(query=goal, project=project, limit=10)
  • Parse into candidate files with relevance scores

5. Load Project Knowledge

  • knowledge_query(project=project, section="patterns.coding")
  • knowledge_query(project=project, section="architecture.key_modules")
  • knowledge_query(project=project, section="gotchas")

6. Build Implementation Plan

Output markdown with:

  • Goal, Issue, Project, Constraints
  • Issue Context: summary, acceptance criteria (if Jira)
  • Files to Modify: checklist from code search
  • Implementation Steps: Setup, Research, Implement, Test, Document, Review, Submit
  • Patterns to Follow: from knowledge
  • Risks & Gotchas: from knowledge or defaults
  • Unknowns to Resolve: placeholder
  • Ready to Start: persona_load('developer'), skill_run('start_work', '{"issue_key": "..."}')

7. Log

  • memory_session_log("Created implementation plan", "Goal: {goal}, Issue: {issue_key}")

Key MCP Tools

  • persona_load — developer persona
  • jira_view_issue — issue context
  • code_search — related files
  • knowledge_query — patterns, architecture, gotchas
  • memory_session_log — session logging

Chaining

  • After: research_topic, gather_context
  • Before: start_work, create_mr, create_jira_issue