From dev-plan
Produce a structured development plan (Markdown) for a feature, refactor, bug fix, migration, new project, or any other coding initiative, and save it under .claude/dev_plans/ in the current project. Use this skill whenever the user asks for a development plan, an implementation plan, a roadmap, a technical plan, a step-by-step plan, or a "how would you build this" — phrases like "piano di sviluppo", "piano di implementazione", "piano tecnico", "fammi un piano per", "preparami un piano", "come implementeresti", "come svilupperesti", "roadmap tecnica", "piano dettagliato", "scrivi un piano per", "implementation plan", "development plan", "technical plan", "technical roadmap", "step-by-step plan", "how would you implement", "draft a plan for", "plan the work for", "outline the implementation". Always use this skill when the user asks for any kind of development/implementation plan, even if they don't explicitly say "piano" or "plan" — intent to plan technical work is enough to trigger it. By default, saves the plan as a Markdown file under .claude/dev_plans/ in the current project.
How this skill is triggered — by the user, by Claude, or both
Slash command
/dev-plan:dev-planThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
When the user asks for a development plan (a feature, refactor, bug fix, migration, new project, etc.), follow this workflow.
When the user asks for a development plan (a feature, refactor, bug fix, migration, new project, etc.), follow this workflow.
Default behavior — unless the user specifies otherwise, save the plan here:
<project-root>/.claude/dev_plans/<descriptive-name>.md
Where <descriptive-name> reflects what is being planned (e.g., auth-rewrite.md, payments-integration.md, migration-postgres-15.md, fix-race-condition-queue.md). Create the .claude/dev_plans/ directory if it doesn't exist.
Override — if the user provides an explicit output path or format (e.g., "save it in docs/", "write it to plan.md", "just show it here", "non serve salvarlo"), use that instead.
find/ls, read key files (entry points, configs, the modules that will be touched). For a brand-new project, skip this step..md file to the default location (or user-specified location), then share a link to it.Use this template as a starting point. Adapt section depth to the size of the initiative — a single bug fix warrants a lighter plan than a full feature or migration.
# Dev Plan: [Name of the initiative]
**Date**: YYYY-MM-DD
**Author**: [user / team]
**Status**: Draft | In review | Approved | In progress | Done
**Target paths**: `path/to/folder`, `path/to/other` (if applicable)
## 1. Goal & context
What we're building or changing, and why. Business motivation, user need, or technical driver
behind the work. Keep it short — 1–2 paragraphs.
## 2. Scope
**In scope**: what this plan covers.
**Out of scope**: what is explicitly NOT covered (so expectations are clear).
## 3. Requirements
Functional and non-functional requirements: what the result must do, performance targets,
security/compliance constraints, accessibility, etc.
## 4. Current state (if applicable)
How things work today and where the gaps/problems are. Cite file paths or modules where relevant.
Skip this section for greenfield work.
## 5. Proposed approach
The chosen architectural/technical direction. Include a brief diagram or component breakdown if
useful. Explain the key design decisions.
### Alternatives considered
If there are meaningful trade-offs, list 1–2 alternatives and why they were rejected.
## 6. Implementation plan
Break the work into ordered phases. Each phase should be independently shippable when possible.
### Phase 1 — [name]
**Goal**: what this phase achieves.
**Tasks**:
- [ ] Task A — short description (touches `path/to/file`)
- [ ] Task B — short description
**Deliverable**: what exists at the end of the phase.
**Estimate**: rough size (e.g., S / M / L, or days).
### Phase 2 — [name]
…
## 7. Risks & mitigations
Things that could go wrong (technical risk, unknowns, external dependencies, data migration
hazards) and how we plan to address them.
## 8. Testing & validation
How we'll verify the work: unit tests, integration tests, manual QA, staging rollout, metrics
to monitor, rollback plan.
## 9. Open questions
Things still to be decided or that need input from someone else. List them so they don't get lost.
## 10. References
Links to related issues, PRs, design docs, prior analyses (e.g., a `.claude/analysis/*.md` file),
external documentation.
.claude/analysis/, reference it in section 10 — it gives the plan grounded context.Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub enesisrl/claude-skills