kunchenguid/vision

Agent skill that mines your repo's history to draft a VISION.md, stress-tests it with hard hypotheticals, and iterates with you on an interactive review board.

What it solves

It solves the problem of undocumented project values and "vibes" that lead to inconsistent decision-making. Instead of relying on memory or generic engineering virtues, it helps project owners codify a testable acceptance policy (a VISION.md file) that clearly defines what a project will and will not become, based on actual historical evidence.

How it works

The tool follows a structured pipeline to generate a project vision:

  1. Evidence Mining: It analyzes the repository's history, specifically merged PRs and commit logs, to find real examples of what was built or declined.
  2. Drafting: It creates an initial VISION.md covering identity, principles, and non-goals.
  3. Stress-Testing: It generates 8-12 "fault-line" hypotheticals—difficult scenarios where project principles might collide—to force the owner to make hard choices.
  4. Interactive Review: The owner uses a specialized review board to provide verdicts on these hypotheticals, and the reasoning behind these answers is folded back into the final document.

Who it’s for

Project maintainers and owners who want to create a rigorous, evidence-based set of contribution criteria for their contributors and AI coding agents.

Highlights

  • Evidence-based: Bans generic virtues; every principle must cite real merged PRs or commits.
  • Steelmanned Hypotheticals: Uses challenging scenarios rather than simple brainstorming to refine the vision.
  • Dedicated Review Surface: Provides a specific stylesheet and board template for a focused, card-based review process.
  • Agent-Ready: Ships as an Agent Skill that can be invoked via slash commands or natural language.

Related

  • Project
    AaravKashyap12/advise-project-approach

    An agent skill that forces AI agents to provide evidence-based project planning and architectural advice by researching real-world comparables and constraints.

  • Project
    shadcn/improve

    An agent skill that audits codebases and generates detailed, self-contained implementation plans for cheaper AI models to execute.

  • Project
    muratcankoylan/Agent-Skills-for-Context-Engineering

    A collection of platform-agnostic Agent Skills focused on context engineering to optimize how AI agents manage their limited attention budget and context windows.

  • Project
    darkrishabh/agent-skills-eval

    A test runner for the Agent Skills ecosystem that empirically measures the performance lift of domain knowledge files by comparing model outputs with and without those skills.

  • Project
    MengTo/Skills

    A Markdown‑based library of 123 "skills"—step‑by‑step, versioned prompt/workflow recipes—for AI coding agents (Codex, Claude, Cursor, etc.). Each skill lives in `agent-skills/<category>/<skill>/SKILL.md` and may include demos, assets, and reference links. The repo lets developers treat prompts as reusable code assets, enabling repeatable UI design, web‑GL, Three.js game, and media‑generation workflows. MIT‑licensed and ready to be loaded as context by any LLM‑driven development tool.