Agent Skills: visual-planning

Use this skill to make a visual plan before implementing a task. Use when prompted to, or suggest to use it when a task is complex enough to require a visual plan.

UncategorizedID: viktorvan/dot-files/visual-planning

Install this agent skill to your local

pnpm dlx add-skill https://github.com/viktorvan/dot-files/tree/HEAD/opencode/skill/visual-planning

Skill Files

Browse the full folder contents for visual-planning.

Download Skill

Loading file tree…

opencode/skill/visual-planning/SKILL.md

Skill Metadata

Name
visual-planning
Description
Use this skill to make a visual plan before implementing a task. Use when prompted to, or suggest to use it when a task is complex enough to require a visual plan.

GOAL Create an implementation-ready plan for the Task using a Mermaid flowchart + a short summary. DO NOT write or suggest any code yet.

WORKFLOW (follow in order)

  1. DISCOVERY PHASE (do this FIRST, before any diagram)

    • Analyze the task requirements thoroughly
    • Explore relevant code/docs/context as needed
    • Identify unknowns, ambiguities, and decisions that need user input
    • Ask ALL clarifying questions in a single message
    • WAIT for user responses before proceeding
  2. PLANNING PHASE (only after discovery is complete)

    • Once you have sufficient clarity, create the visual plan
    • Output EXACTLY two parts: the Mermaid diagram + 3-5 bullet summary
    • The bullets summarize the COMPLETED plan, they do NOT ask questions

HARD RULES

  • Do NOT create the diagram until all critical questions are answered
  • Do NOT implement, draft code, propose patches, or include file contents. Planning only.
  • Do NOT begin implementation until I explicitly reply with: "APPROVED: IMPLEMENT".
  • The final plan output must contain EXACTLY two parts and NOTHING ELSE: (1) a single Mermaid code block (2) exactly 3–5 bullet points summarizing the plan
  • No preface, no extra headings, no extra paragraphs, no checklists, no "next steps" outside the bullets.
  • Bullets are a SUMMARY of decisions made, NOT questions or open items.

MERMAID REQUIREMENTS

  • Use: flowchart TD
  • DO NOT use quoted labels (no ["..."] or subgraph X["..."])
  • All node and subgraph labels MUST be unquoted plain text
  • Start node label: Start: <brief task summary> (no quotes)
  • End node label: End: <expected outcome> (no quotes)
  • Use shapes:
    • Actions/steps: [like this]
    • Decisions/conditions: {like this}
    • Inputs/outputs: (like this)
  • Every decision must have labeled edges (yes/no or success/error).
  • Include: validation, edge cases, failure handling, and rollback/escape path where relevant.
  • Include integration points (APIs/services/modules), test strategy nodes, and observability (logging/metrics) if applicable.
  • Keep nodes implementation-oriented (what to do), not narrative.

BULLET REQUIREMENTS (exactly 3–5 bullets)

  • Bullets summarize: key design decisions, approach chosen, and any accepted tradeoffs.
  • These are statements about the plan, NOT questions or requests for confirmation.
  • No additional formatting besides simple "-" bullets.