nick-vels/skills

Describe what you want built — get a finished project. An agent skill for Claude Code, Cursor and Codex: asks only where your idea forks, then writes the spec, splits it into tasks, builds it and checks the result against your original brief. Live progress on a dashboard that opens by itself.

What it solves

Autopilot is a development framework designed to take a high-level idea and turn it into a fully functional software project without requiring the user to write code, read technical specifications, or manage tasks. It bridges the gap between a vague concept and a production-ready application by automating the specification, planning, and execution phases of development.

How it works

Autopilot operates as a "skill" for AI agents (like Claude Code, Cursor, or Codex). It follows a structured multi-phase pipeline:

  1. Preparation & Requirements: It converts the user's natural language description into a numbered list of verbatim requirements.
  2. Briefing: The agent asks targeted questions to resolve ambiguities and define critical infrastructure (e.g., payment gateways, hosting).
  3. Specification & Planning: Requirements are expanded into detailed scenarios and then broken down into a set of tasks organized into "waves" for parallel execution.
  4. Development: Each task is handled by a separate agent with a clean memory to prevent context drift. Independent tasks are developed in parallel, and each task results in a single commit.
  5. Review & Acceptance: The project undergoes code review and a "blind acceptance" test where a separate agent verifies the final product against the original user requirements without seeing the intermediate specifications.

Throughout the process, it provides a real-time HTML dashboard that tracks requirements, task progress, and system status.

Who it’s for

  • Non-programmers who want to build a project "turnkey" from a text description.
  • Users who prefer a high-level overview rather than line-by-line coding assistance.
  • Project owners who want to ensure every single original requirement is implemented and verified.

Highlights

  • Requirement Tracking: Every user requirement is tracked verbatim; nothing is silently dropped during the transition from brief to code.
  • Dynamic Control: Offers four modes (Full Auto, Semi-Auto, Interview, Manual) and three levels of depth (Strict, Normal, Deep) to control how much the AI assumes versus asks.
  • Isolated Task Execution: Uses separate agents for each task to avoid context pollution and ensure stability.
  • Self-Documenting Projects: Automatically generates AGENTS.md and architecture docs so future AI sessions can understand the project instantly.
  • Blind Acceptance: A final verification step where an independent agent checks the result against the original brief to ensure fidelity.

Related

  • Project
    aiblueprinthq/ai-blueprint

    AI Blueprint is a file‑backed workflow framework that lets LLM coding agents (Codex, Claude Code, Copilot, etc.) develop features in a disciplined, spec‑first manner. It stores specs, audit findings, and history as markdown files in the repo, provides a set of chat‑driven commands (/feature, /implement, /check, /audit, /complete), and includes a local dashboard for visibility. The tool integrates with Git, supports parallel work via separate worktrees, and can generate CI checks. It is distributed via npm, MIT‑licensed, and documented on ai‑blueprint.dev.

  • Project
    github/spec-kit

    Spec Kit is an open-source toolkit that provides structured, repeatable processes for AI coding agents to handle feature development, bug fixing, and project assessment.

  • Project
    Significant-Gravitas/AutoGPT

    AutoGPT is an AI-powered platform that lets you manage a team of specialized agents to automate recurring professional tasks like content creation, research, and administrative work.

  • Project
    growthxai/output

    Output is an open‑source TypeScript framework that lets you define, run, and monitor LLM‑powered workflows. It bundles prompt files, multi‑provider LLM access, automatic tracing, cost tracking, evaluators, and Temporal‑based orchestration, all under one npm package suite.

  • Project
    qf-studio/pilot

    An AI software engineer that automatically picks up tickets from GitHub, Jira, or Linear, implements the code, and opens pull requests for review.