plan · codewithmukesh — Claude Code skill for .NET planning
Builds an architecture-aware implementation plan for .NET work before code changes begin. Use it for changes spanning layers, modules, or architectural decisions.
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a claude-codeWhat plan does
This skill switches the agent into planning mode before implementation. It checks docs/specs/ for an approved specification, identifies the project architecture or recommends one, then maps the task to affected layers, modules, and boundaries.
It uses the architecture-advisor skill for architecture detection and can use the find_references and find_callers MCP tools to examine change impact. The output is a numbered plan with affected layers, open questions, and risks, followed by a request for approval. It is excessive for a one-file edit or a simple bug fix.
When a feature cannot fit into one sentence and no specification exists, plan recommends /spec before planning. If implementation goes off track, it calls for revising the plan rather than continuing with a broken approach.
What it helps with
- plan a feature that spans API and domain layers
- identify the architecture pattern in a .NET project
- map files, layers, and modules affected by a change
- check change impact through references and callers
- revise an implementation plan after an issue appears
Example prompts for the agent
/plan Design a product catalog with search and filtering/plan Think through a refactor that affects multiple consumers/plan How should I implement a feature across API, application, and infrastructure?
How to install plan
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a claude-code2Manually: put the files in place
~/.claude/skills/plan/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/plan/
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a codex2Manually: put the files in place
.agents/skills/plan/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a cursor2Manually: put the files in place
.cursor/skills/plan/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install codewithmukesh/dotnet-claude-kit plan2Manually: put the files in place
.github/skills/plan/SKILL.md1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a gemini-cli2Platform CLI
gemini skills install https://github.com/codewithmukesh/dotnet-claude-kit --path skills/planManually: put the files in place
.devin/skills/plan/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about plan
How do I install plan in Claude Code?
Run npx skills add codewithmukesh/dotnet-claude-kit --skill plan -a claude-code. The skill file is placed at ~/.claude/skills/plan/SKILL.md.
What does the plan skill do?
It examines a .NET task through the project architecture and produces a numbered implementation plan before code is written. The plan covers affected layers, open questions, and risks.
When does plan recommend creating a spec first?
When a feature is too large to describe in one sentence and no approved spec exists in docs/specs/, it recommends /spec. An approved spec drives the planning steps.
Source text
SKILL.md · MIT
--- name: plan description: > Enter plan mode for .NET projects with architecture awareness. Analyzes tasks through the lens of supported architectures (VSA, Clean Architecture, DDD, Modular Monolith) and produces structured implementation plans before any code is written. Use when: "plan", "let's plan", "think through", "design this", "how should I implement", or any non-trivial task requiring 3+ steps. --- # /plan -- Architecture-Aware Planning ## What Enters a structured planning mode that considers the project's architecture pattern before producing an implementation plan. Instead of jumping straight to code, this command forces a deliberate pause to: - Identify the project's current architecture (or recommend one) - Map the task to affected layers, modules, and boundaries - Produce a numbered implementation plan with clear steps - Iterate on the plan until it is solid before writing any code Plans are living documents -- if something goes sideways during implementation, stop and re-plan rather than pushing through a broken approach. ## When - Non-trivial tasks requiring 3 or more implementation steps - Tasks involving architectural decisions (new modules, cross-cutting concerns, new bounded contexts) - Features that touch multiple layers (API, application, domain, infrastructure) - Refactoring that could affect multiple consumers - Any time the user says "plan", "think through", "design this", or "how should I approach" **Skip planning for:** Single-file changes, simple bug fixes, typo corrections, config tweaks. ## How ### Step 1: Understand the Task Check `docs/specs/` for an approved spec covering this task — if one exists, it is the source of truth: map its acceptance criteria to plan steps and skip ahead to Step 2. If the feature is spec-worthy (too big to describe in one sentence) and no spec exists, recommend `/spec` first instead of planning on assumptions. Otherwise, clarify what the user wants to build with focused questions. Do not assume requirements that were not stated. ### Step 2: Detect Architecture Use the `architecture-advisor` skill to determine the project's architecture: - Check for existing architecture markers (folder structure, project references, patterns) - If no architecture is established, run the architecture questionnaire - Load the appropriate architecture-specific skill (vertical-slice, clean-architecture, ddd) ### Step 3: Map Affected Areas Identify every layer, module, and boundary the task touches: - Which projects/folders will have new or modified files? - Are there cross-cutting concerns (auth, caching, validation, logging)? - What existing code will be impacted? Use `find_references` and `find_callers` MCP tools for blast radius. - Are there database migrations needed? ### Step 4: Produce the Plan Output a numbered plan with this structure: ``` ## Plan: [Task Title] **Architecture:** [Detected architecture] **Affected layers:** [List] **Estimated steps:** [Count] ### Steps 1. [Step] -- [Which file/layer] -- [Why this order] 2. ... ### Open Questions - [Anything that needs user input before proceeding] ### Risks - [Potential issues and mitigations] ``` ### Step 5: Iterate Present the plan and ask: "Does this plan look right, or should I adjust anything?" Revise until the user confirms. Only then proceed to implementation. ## Example ``` User: /plan Add a product catalog feature with search and filtering Claude: Let me analyze your project structure first... ## Plan: Product Catalog Feature **Architecture:** Vertical Slice Architecture **Affected layers:** Features/, Infrastructure/Persistence/, Tests/ **Estimated steps:** 6 ### Steps 1. Create Product entity + EF config -- Domain layer -- Foundation for everything else 2. Add migration -- Infrastructure -- Schema must exist before features 3. Scaffold GetProducts feature (with filtering) -- Features/Products/ -- Read path first 4. Scaffold CreateProduct feature -- Features/Products/ -- Write path 5. Add integration tests -- Tests/Features/Products/ -- Verify both features 6. Wire up OpenAPI metadata -- Features/Products/ -- Documentation ### Open Questions - Should search be full-text (PostgreSQL tsvector) or simple LIKE? - Do products need categories/tags for filtering? Does this plan look right, or should I adjust anything? ``` ## Related - `/spec` -- Write the spec first for features too big to describe in one sentence - `/scaffold` -- Generate the files once the plan is approved - `/verify` -- Run verification after implementing the plan
© codewithmukesh · MIT · codewithmukesh/dotnet-claude-kit
Similar skills
All skills for Agent setup: memory, context, orchestration →plan by other authors comes first.