SuperAPI AI Directory

Blueprint

Open-source planning copilot for coding agents. Asks multiple-choice questions about your task, then generates a markdown plan any coding agent can execute. Extensions for VS Code, Cursor, and Windsurf.

Open website

About this tool

Blueprint is a planning copilot tool for coding agents, launched by Imbue. Its core idea is: ask the right questions before writing code, then hand the agent a plan it can execute in one go. Most coding agents rush to write code or make a plan based on guesses, while Blueprint deliberately slows down and first understands the task clearly.

Uses

  • Clarify task requirements and implementation direction before starting to code.
  • Turn ideas in your head into specifications and an executable plan.
  • Help uncover details you may not have thought of or considered.

Main capabilities

  • Read your codebase as the basis for planning.
  • Ask multiple-choice questions that are easy to answer, and you can skip the parts you do not care about.
  • Once the questions cover enough, generate a plan.
  • Output a Markdown plan that any coding agent can execute.
  • Built-in multiple templates. The default template includes: Overview, Expected Behavior, Implementation Plan, Implementation Phases, Testing Strategy, Open Questions. The concise template includes: Overview, Expected Behavior, Changes.
  • Supports describing custom templates inline during the onboarding flow, and templates can also be managed in settings.
  • The plan is written to blueprint/<slug>/plan.md.
  • In addition to the VS Code extension, it also provides agent-agnostic skills for Claude Code, Codex CLI, Gemini CLI, and other compatible carriers, see imbue-ai/blueprint.

How to use

  1. Install Blueprint from the VS Code Marketplace.
  2. Click the Blueprint icon in the activity bar to open the sidebar.
  3. Describe your task.
  4. Answer the questions Blueprint asks; skip the ones you do not care about.
  5. Generate a plan once enough information is covered.
  6. Continue refining it in the conversation, or hand the plan to your coding agent.

Suitable scenarios

  • Brand-new projects (greenfield projects).
  • Developing large new features on an existing codebase.
  • Incremental changes that are worth planning first.
  • Exploratory experiments.
  • New models, new systems, or new subsystems.

Not ideal scenarios

  • Frontend work where most decisions depend on visual judgment.
  • Small refactors.
  • Debugging and polishing work.

Platform support Can be used in VS Code, Cursor, and Windsurf.

Limitations Apart from the information above, details about pricing, cost, platform specifics, etc. have not been provided.