pradeepmouli/speckit-refine-propagate

Refine Propagate Skill

Propagate Spec Changes

After refining a specification with /speckit.refine.update, use this command to cascade changes downstream to plan.md and tasks.md without regenerating them from scratch.

User Input

$ARGUMENTS

You MUST consider the user input before proceeding (if not empty). The user may specify which artifacts to propagate to (e.g., "only plan" or "plan and tasks").

Prerequisites

1. Verify a spec-kit project exists by checking for .specify/ directory
2. Locate the current feature's spec directory (same logic as /speckit.refine.update)
3. Verify spec.md exists and has been refined (check for **Refined**: entries or **Status**: Refined)
4. Check which downstream artifacts exist: plan.md, tasks.md

Outline

1. Load artifacts: Read from the current feature directory:

   • Required: spec.md (the refined specification)
   • Target: plan.md and/or tasks.md (artifacts to update)
   • Optional: research.md, data-model.md, contracts/ (for context)

2. Identify what changed: Parse the refinement notes in spec.md:

   • Look for **Refined**: entries to understand the change history
   • Look for ~~removed~~ items to identify dropped requirements
   • Compare user stories, requirements, and success criteria against plan.md/tasks.md

3. Propagate to plan.md (if it exists and user didn't exclude it):

   • Added user stories: Add corresponding sections to the plan (project structure, complexity tracking)
   • Modified requirements: Update the Technical Context, Project Structure, or Complexity Tracking sections as needed
   • Removed requirements: Mark corresponding plan sections with strikethrough and reason
   • Preserve: All existing plan structure, formatting, and sections not affected by the change
   • Remove any ⚠️ **STALE** warning from the top of the file
   • Add a propagation note:

     **Propagated**: [DATE] — Updated from spec.md refinement
     

4. Propagate to tasks.md (if it exists and user didn't exclude it):

   • Added user stories: Add a new Phase section with tasks following the existing format ([ID] [P?] [Story?] Description [(depends on ...)])
   • Modified requirements: Update affected tasks in-place — adjust descriptions, dependencies, or file paths
   • Removed requirements: Mark affected tasks with ~~[REMOVED]~~ prefix and strikethrough, do not delete
   • Renumber if needed: If new tasks are added, assign IDs continuing from the last used ID
   • Update Execution Wave DAG: If the tasks.md has a Wave DAG section, regenerate it to reflect the new task dependencies
   • Update Dependencies section: Adjust phase dependencies and user story dependencies
   • Remove any ⚠️ **STALE** warning from the top of the file
   • Add a propagation note:

     **Propagated**: [DATE] — Updated from spec.md refinement
     

5. Validate consistency: After propagation, verify:

   • Every user story in spec.md has a corresponding phase in tasks.md
   • Every requirement in spec.md is traceable to at least one task
   • No orphaned tasks reference removed requirements
   • Dependencies still form a valid DAG (no circular dependencies)

6. Report: Output a summary:

   • What changed in each artifact
   • How many tasks were added, modified, or marked removed
   • Any consistency warnings found during validation
   • Suggest next step: /speckit.refine.diff to review changes, or /speckit.implement to continue building

Rules

• Never regenerate from scratch — propagate surgically updates only affected sections
• Never delete content — mark removed items with strikethrough
• Preserve all formatting — match existing artifact styles
• Track changes — always add propagation notes with dates
• Remove staleness warnings — after successful propagation, clear the stale markers
• Maintain traceability — every spec requirement must map to plan/task items