sync-vision

Sync root VISION.md from changed project docs, plans, and rules.

1.2k|81|Updated Sep 5, 2024
One-click install
npx skills add https://github.com/udecode/dotai --skill sync-vision
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: sync-vision
Source: https://github.com/udecode/dotai/tree/main/skills/sync-vision
Command: npx skills add https://github.com/udecode/dotai --skill sync-vision

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Project doctrine scattered across plans, docs, skills, and reviews drifts out of sync with the root VISION.md file, causing agents and maintainers to repeat past mistakes. This Skill incrementally classifies changed inputs and patches VISION.md with reusable project taste.

Core Features & Use Cases

  • Incremental Diff Collection: A helper script collects changed durable inputs since the last synced commit, including working-tree overlays, and extracts candidate doctrine lines by category.
  • Classification Ledger: Every candidate is classified as captured, reaffirmed, rejected, run-specific, owner-routed, or deferred-with-question, with baseline advancement only after full accounting.
  • Owner Routing: Non-vision misses are routed to the owning skill, spec, research doc, or migration guide instead of polluting VISION.md.
  • Use Case: After a week of merged plans and rule changes, run the sync command to patch VISION.md with new reusable doctrine and record a decision ledger in the run artifacts.

Quick Start

Ask the agent to run sync-vision to analyze changed inputs since the last synced commit and update the root VISION.md with any new reusable project doctrine.

Frequently Asked Questions about sync-vision

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I keep a project VISION.md file up to date automatically?

Run the sync-vision workflow, which diffs committed inputs since the last synced commit, extracts candidate doctrine lines, and patches VISION.md only for reusable decisions. The helper script writes changed-files, candidate-lines, and summary artifacts for review.

What inputs does the vision sync analyze?

It analyzes VISION.md, AGENTS.md, .agents/rules, skills, docs/plans, docs/research, docs/solutions, content/docs, and other changed Markdown-like root docs. Generated .agents/skills mirrors are excluded as primary inputs, and extra paths can be added via docs/sync/vision/config.json.

Can I preview vision changes without modifying VISION.md?

Yes, run the helper with the preview or dry-run mode to write run artifacts and recommendations without patching VISION.md or advancing the baseline. The status mode only reports counts with no writes at all.

When does the sync baseline advance to the latest commit?

The baseline advances only when every candidate cluster is classified, VISION.md is patched or reaffirmed, owner-routed items have owners, and the active plan passes its completion check. Preview runs, unresolved taste questions, or failed mirror syncs block advancement.

Does sync-vision require other skills or tools?

It requires git and Node.js to run the collect-vision-diff.mjs helper, and depends on the autogoal skill for goal lifecycle, plan templates, and completion checking. A repo-owned docs/plans/templates/sync-vision.md template is used when present.