Type to search pages and rules.

4.0.0

Use with Agents

An AI coding agent follows LiteSpec only when it reads the rules before it writes code. Every LiteSpec page is published as plain Markdown that an agent can fetch. This page sets that up: read how LiteSpec works yourself, start one session with a prompt, or answer four questions about your project and commit the block they produce to its AGENTS.md file, which agents read at the start of every session.

  1. Read How It Works

    The agent applies the rules, but you decide what it may do on its own and you review what it brings back. These 10 pages explain the rules and how LiteSpec's parts fit together, in reading order. They take about 73 minutes in all.

    1. Why LiteSpec Exists5 min

      The problem LiteSpec solves, for a team that is deciding whether LiteSpec fits.

    2. How LiteSpec Works7 min

      The standard, the records, the model, and the tools, traced through one Claimwick use case.

    3. Read the Standard5 min

      Project selections, page arrangement, and provision IDs; how to find the rules that apply to a project.

    4. Records and the Project Model10 min

      The four layers, what a project writes down, where each record lives, and what the model builder computes.

    5. Reading Tracks6 min

      Which records to read in which order: why before how, one journey end to end, one module end to end, and the structural words.

    6. How the Tools Fit Together8 min

      The CLI, the analyzers, and the site, in the order one change moves through them.

    7. How LiteSpec Compares15 min

      LiteSpec beside ten other approaches to spec-driven development, and when each fits.

    8. Lineage4 min

      The practices LiteSpec borrows from, the words it takes from each, and what it does differently.

    9. Work with Agents4 min

      How an agent finds the rules, what it decides alone, and what it brings back.

    10. Adopt LiteSpec in Steps9 min

      Adopting LiteSpec by conformance level: the record model at `records`, traceability at `traced`, and the full contract at `full`, with the records path through the tutorial.

  2. Start a Session with One Prompt

    Paste this prompt into an agent session to try LiteSpec without setting anything up. The agent reads the site index, picks the selection index that matches your project's shape and stacks, and fetches only the pages each task needs. The setup lasts for that session only.

    Prompt for Your Agent

    Read https://litespec.dev/llms.txt. Pick the selection index that matches this project's shape and stacks,
then fetch only the pages each task touches.

  3. Tailor It to Your Project

    For a repository you keep working in. Your answers pick a selection index: a Markdown file,llms.txt, that lists only the LiteSpec pages your project applies, so the agent never loads rules for a stack the project does not use.

    PresetFills in the shape and stacks. Change them after.
    SizeSmall: one team, no journeys or scenarios. Large adds journeys, scenarios, and the Acceptance project.
    ShapeServer: one API deployable and one database. Client-only: no server of its own.
    Backend stackOne backend stack for a server project.
    Frontend stackOne stack per frontend, or none. A stack that no published selection pairs with your answers is disabled.
    ExtensionsEach applies only when you select it. Only extensions that support your stacks are listed.

  4. Copy It into Your Project

    Paste this block into the AGENTS.md file at the root of your repository and commit it. An agent that reads AGENTS.md at the start of a session then loads your selection index before each task.

    AGENTS.md Block

    ## LiteSpec
This project follows LiteSpec 4.0.0: size small, shape server, stacks dotnet and react, extensions none.
Before a task, read https://litespec.dev/4.0.0/select/server.dotnet.react/llms.txt and fetch only the pages the task touches.
Selected extensions: none
UI work: the pattern pages listed in that index.