/atn-mindmap: shape a specification by answering questions

/atn-mindmap runs a mind map on a subject you name and leaves one Prepared specification in your project. Run it for every piece of work you want thought through before code is written, from a one-line change to a whole application.

A mind map is a session in which your AI agent asks rounds of questions and turns your answers into a specification. A specification is one described piece of work, with a title, a summary, Markdown content, a status and its own questions and answers. The workflow page shows where the mind map sits among the other skills.

Run /atn-mindmap

/atn-mindmap [subject] [--continue <spec-id> | --from-todo <card-id>] [--auto-apply] [--auto-decompose]

Argument or flag	Description
subject	Optional free text that says what the specification is about. Without it, the skill takes the subject from the conversation, or asks for it.
--continue <spec-id>	Continues the mind map on an existing specification and reuses its recorded questions and answers. Cannot be combined with --from-todo.
--from-todo <card-id>	Starts the mind map from a ToDo card. The card's title and description are the subject, and text you add only narrows or steers it. Cannot be combined with --continue.
--auto-apply	Skips the question about the next step and applies the specification.
--auto-decompose	Skips the question about the next step and decomposes the specification into the stages the skill proposes. Together with --auto-apply, the stages are applied too.

The two automation flags combine with each other and with --continue or --from-todo, in any order. They change the next step only, and every other question is asked. If you pass both --continue and --from-todo, the skill says so and stops without creating or changing anything.

/atn-mindmap Guests can check out without creating an account

In the web app, a Draft specification offers Copy mind-map command, which copies the --continue form with the id filled in. An open ToDo card that has no specification offers the same label for the --from-todo form.

What /atn-mindmap does

1. Takes the subject from your command or from the conversation. If neither holds one, it asks what the specification is about.
2. Researches the subject before the first question: where the specification belongs in the specification tree, whether it is feasible in your codebase, and which existing specifications it depends on, overlaps or conflicts with. The research uses search by meaning, so it finds related specifications that are worded differently.
3. Asks you to confirm the placement.
4. Creates the specification as a Draft under the parent you confirmed, with a working title. From here on, each question and your answer are saved as you give them, and the specification's Questions & answers tab in the web app fills as you go.
5. Runs the question rounds. A round covers one topic with 3 to 8 questions, each with 2 to 5 options, on matters such as scope, constraints, edge cases, users, integration points and what is left out. The first round draws on what the research found in your codebase.
6. Asks for the next topic after each round, offering three topics that are not covered and "Finish the mind map". You steer the session by picking the topic, and you end it when the specification is complete enough for you.
7. Writes the final title, summary and content from all the rounds, organised by topic. When a link or a local path you gave, or web research the skill did, informed a decision, the content ends with a ## References section that lists those sources.
8. Sets the specification's blockers from the research, without asking you.
9. Sets the specification to Prepared.
10. Reports, then asks what to do next.

The specification tree is the hierarchy of a project's specifications, and a blocker is a dependency between two specifications: a specification that is blocked by another is implemented after it.

A small change needs only the mind map

A mind map is worth running on a change that looks too small to specify, because the questions bring out what you have not decided. Take a web shop whose product page has the tabs "Description" and "Reviews":

/atn-mindmap Put the Reviews tab before the Description tab on the product page

One round on such a subject asks whether "Reviews" becomes the tab that opens by default, whether its content loads with the page or when the tab is opened, what a product with no reviews shows, and whether existing links to a tab keep opening that tab. You answer, choose "Finish the mind map" and then "Apply now".

The specification has no sub-specifications, so it is a leaf specification, and /atn-apply plans and implements it as it is. Decomposing is for a specification that covers more than an agent should build in one go.

Continue an existing specification

/atn-mindmap --continue <spec-id> picks up a specification that exists, whether a mind map created it or you wrote it by hand in the web app. The skill keeps the specification where it is in the tree, researches feasibility and dependencies again, and reads the recorded questions and answers so that it does not ask them twice. If the specification has recorded questions, the skill first asks you to confirm or clarify the subject.

The rounds then run as for a specification that starts from nothing, and the status stays as it was until the end. If the session changed the direction of the specification, the skill rewrites the content from scratch. Otherwise it keeps the content and merges the added material into it. It then sets the blockers and sets the specification to Prepared, whatever its status was before.

Start from a ToDo card

A ToDo card is a short note of work to do later: a follow-up, tech debt, a bug or an idea. /atn-mindmap --from-todo <card-id> turns a ToDo card into a specification:

1. The card's title and description become the subject.
2. Before the first round, the skill asks each open question of the card that has no answer, with the options written in the question, and stores each answer on the card.
3. Every question of the card, with its answer, is copied into the specification's questions and answers, so the finished specification takes them into account.
4. The specification is linked to the card when it is created, and the card shows it under Became.
5. When the specification is Prepared, the skill sets the card to Done.

What the skill does depends on the state of the card:

The card	What happens
Open, with no specification	A specification is created and linked to the card.
Open, with a specification in Draft	The skill continues that specification. A card has at most one specification.
Done or Dismissed with a specification, or its specification is past Draft	The skill names the card's status and the specification's status and asks whether to continue.
Done or Dismissed, with no specification	Nothing is created. Reopen the card in the web app's archive first.

Questions /atn-mindmap asks

When	The question and its options
No subject was given and the conversation holds none	What the specification is about, as an open question
After the research	Whether the recommended place in the tree is right: "Yes, place it there", "No, top-level instead" or "Let me specify a different parent". For a different parent, the skill lists the top-level specifications, and you choose one or type an id.
In each round	3 to 8 questions on the topic, each with 2 to 5 options
After each round	The next topic: 3 candidate topics, or "Finish the mind map"
At the end	The next step: "Apply now", "Decompose into stages" or "Nothing else"
After "Decompose into stages"	The skill shows the proposed stages with a one-line scope each and asks "Yes, create these as drafts" or "No, skip decomposition for now"
With --continue, when the id is not valid	Whether to give a corrected id or start a specification from nothing
With --continue, when the specification has sub-specifications	A warning, and whether to continue anyway
With --from-todo, when the card id is not valid	Whether to give a corrected card id or start a specification without a card
With --from-todo, when the card is closed and has a specification, or its specification is past Draft	"Continue the existing specification" or "Stop"

With --auto-apply or --auto-decompose the question at the end is skipped. With --auto-decompose the proposed stages are also created without the confirmation. The warning about sub-specifications and the question about a closed card are asked under the automation flags too.

"Apply now" starts /atn-apply on the specification, and "Decompose into stages" starts /atn-decompose with the stages you confirmed. Each of them reports in the same session.

What /atn-mindmap changes

Where	What changes
The specification	Created as a Draft at the start, or the existing one with --continue. Its title, summary and content are written at the end, and its status becomes Prepared.
Questions and answers	Each question of each round is stored on the specification with your answer, as you give it.
Blockers	The specification's own blockers are added, and with --continue blockers that the research rejects are removed. The skill does not make another specification blocked by this one.
The ToDo card, with --from-todo	Answers to its open questions are stored, the specification appears under Became, and the card is set to Done. The skill makes no other change to a card.
Your repository	Nothing. A mind map reads your code and writes no file, branch or commit.

A blocker is left out when it points at a Completed specification, at the specification itself or at one of its ancestors or sub-specifications. A blocker that Athenode refuses, such as one that would close a cycle, is also left out, and the mind map carries on.

When your methodology setting is tests first, the content also holds an ## Acceptance criteria section wherever testable criteria can be drawn from your answers. /atn-settings is where you choose the methodology.

The specification and its questions appear in the web app while the session runs, with no reload.

The report

The report states the specification's id. It lists the blockers added, removed and skipped, with the reason for each skipped one.

For a run started from a ToDo card, the report adds the card's id and whether the card was set to Done, or why that was refused. It also lists the card questions that were asked and any answer that could not be stored on the card.

When the next step ran another skill, that skill's own report follows.