Help us improve
Share bugs, ideas, or general feedback.
From piv-speckit
Generates feature implementation plan as split artifacts: spec.md (requirements/workflows), plan.md (technical approach), tasks.md (step-by-step), quickstart.md (TL;DR) in .claude/specs/{feature}/.
npx claudepluginhub galando/piv-speckit --plugin pivHow this command is triggered — by the user, by Claude, or both
Slash command
/piv-speckit:plan-feature <feature-name>commands/piv_loop/The summary Claude sees in its command listing — used to decide when to auto-load this command
# Plan a New Feature **Goal:** Transform feature request into implementation plan with **split spec artifacts**. ## Usage ## Feature: $ARGUMENTS ## Execution > **Full methodology:** `$CLAUDE_PLUGIN_ROOT/.claude-plugin/reference/execution/plan-feature.md` ### Planning Phases - **Phase 0: Auto-Prime** (automatic — loads/refreshes codebase context silently) - Phase 1: Feature Understanding - Phase 1.5: Pragmatic Programmer Principles Review - Phase 2: Codebase Intelligence Gathering - Phase 3: External Research & Documentation - Phase 4: Deep Strategic Thinking - Phase 5: Plan Structur...
/archStarts collaborative feature specification process: investigates codebase for patterns and constraints, proposes design with reasoning and file paths, resolves ambiguities via conversation, saves Markdown spec with behavior, architecture, constraints, and related files.
/featureCreates detailed implementation spec in .claude/specs/*.md for given feature-description after deep codebase analysis.
/planGenerates an implementation plan for a feature by reading its spec, detecting stack (Next.js/Kotlin/etc.), designing architecture/files/tasks, and saving to .planning/plans/<feature>.md.
/plan-featureDefines the minimal sufficient pipeline for a feature, using current repo audit and orchestrating cross-functional roles (PO, UI/UX, Backend, Frontend).
Share bugs, ideas, or general feedback.
Goal: Transform feature request into implementation plan with split spec artifacts.
/piv-speckit:plan-feature "feature description"
Full methodology:
$CLAUDE_PLUGIN_ROOT/.claude-plugin/reference/execution/plan-feature.md
Phase 0: Check .claude/agents/context/prime-context.md. If missing/stale (>20 files changed or >7 days), re-scan codebase silently.
Phase 1: Extract problem, identify type (New/Enhancement/Refactor/Fix), assess complexity (Low/Med/High), map affected systems. Create user story: As a <user> I want <goal> So that <benefit>.
Phase 1.5: Review DRY (reuse patterns?), Broken Windows (tech debt?), Automate (tests/CI?), Design for Change (flexible?).
Phase 2: Read prime context, analyze project structure, find similar implementations, check patterns in .claude/rules/, identify integration points.
Phase 3: Research latest library docs, find examples, identify gotchas.
Phase 4: Think about architecture fit, dependencies, edge cases, testing strategy, performance, security.
Phase 5: Generate plan with: context references, patterns to follow, implementation phases, step-by-step tasks with VALIDATE commands.
Phase 6: Create split artifacts in .claude/specs/{XXX-feature}/ or single file in .claude/agents/plans/.
Split Artifacts (Inspired by GitHub's Spec-Kit)
Creates .claude/specs/{XXX-feature-name}/ with:
spec.md - WHAT: User stories, requirements, workflowsplan.md - HOW: Technical approach, architecture, APIstasks.md - DO: Step-by-step implementation tasksquickstart.md - TL;DR: Quick reference for humansBackward Compatible: