Skip to content
authorship.dev

← Back to the page

The plan in detail

AI doesn't end developer authorship. It separates authorship from making.

If you write only 10 % of the code yourself, when is it still your work?

  1. 1 · Issue

    Intent - Own the why.

    Do I have the why, and have I set what "done" means?

    What it covers

    • The problem: which one, for whom, and is it the right one?
    • A picture of success: what "done" looks like in the result
    • Non-goals and domain limits
    • Risk level

    Anti-pattern"Agent, implement X." Passing the ticket through.

    Core practicesOne sentence on what done means, before you type the first prompt (Opening Line).

  2. 2 · WIP

    Orchestration - Shape the making.

    Have I shaped how it comes into being?

    What it covers

    • The split: what you do yourself, what you delegate, and to whom
    • Context
    • Working limits: permissions, area, diff size, checkpoints
    • Attention: what you do between turns

    Anti-patternFire and forget.

    Core practicesDecide what you delegate, keep and grow; no second hard task between agent runs (Delegation Matrix, Between Turns).

  3. 3 · PR

    Verification - Set the proof.

    What criteria do I check it against, up front?

    What it covers

    • Criteria set in advance: tests, properties, limits, architecture rules
    • Evidence in proportion to the risk
    • A counter-check

    Anti-patternCriteria only after the output. "Tests green" as the only criterion.

    Core practicesSet the criterion before you see the output. Tests first: you set the intent, the AI writes the tests, a test run decides, not the impression (Prophylaxis).

  4. 4 · Sign-off

    Judgment - Say yes deliberately.

    Do I agree because it is right and it fits?

    What it covers

    • Is it right (verification met) and does it fit (context, users, trade-offs)?
    • "No" is a full result

    Anti-patternSigning off out of habit (automation bias).

    Core practicesName one place where this could be wrong, then check it. If you can't, don't sign off (Blunder Check, the Pre-Merge Routine).

    Where do you sign off in your workflow? →

  5. 5 · Production

    Accountability - Stand behind it.

    Do I stand behind it when it breaks?

    What it covers

    • Explain it, repair it, learn from it
    • Blameless

    Anti-pattern"The AI did that."

    Core practicesLook back briefly in the evening, sober and factual, like a postmortem. What you take from it is learning for the next Intent ("Post-Game Analysis").

  6. always

    Craft - Keep your craft.

    Do I understand it well enough to explain and change it?

    What it covers

    • Read, debug, break down and specify code
    • Keep your skills on purpose, and grow them

    Anti-pattern"I can't do this without Copilot anymore."

    Core practicesYour own hypothesis before the prompt; every week 30 to 45 minutes on real work without AI. Classic exercise: the seeded bug (Calculate First, Game Day).

  7. center

    Authorship - What arises when the cycle is run through.

    Who signs here - and for what?

    Core practicesOne sentence on what you are the author of: the README of your role. You read it every week and revise it every quarter (Author Statement).

The cycle repeats: from Production, the "Post-Game Analysis" leads to the next Intent. How deep you go at each station depends on the risk of the change.

Three ways to start →