SuperAPI AI Directory

leaf-idea

Agents & models

View repository

Updated 10-15 22:14

About this skill

LEAF Idea

leaf-idea owns Learn: ① Intent and ② Unknowns & Context. It turns a rough idea into a sprout whose why, provisional what, unknowns, facts, assumptions, and split decision are clear enough for ③ Criteria.

Core Model

  • A sprout is one possible future leaf, not an inbox item.
  • Learn runs in the same sprout that later continues to ③; do not require a stage move.
  • leaf-work takes over at ③ Example after Learn passes, still in the same sprout.
  • Concepts, taxonomies, models, policies, decision records, plans, documents, UI, and code changes can all be valid artifacts. If implementation becomes a separate lifecycle, reviewer, success check, or code surface, split it into a separate sprout.
  • Older compatibility folders may exist as storage data only.

First Read

Inspect local truth before asking:

git status --short --branch
find .leaf/01-sprouts .leaf/02-leaves .leaf/03-fallen -maxdepth 1 -mindepth 1 -type d 2>/dev/null | sort

Resume a likely matching sprout instead of creating a duplicate. Use lowercase ASCII kebab-case slugs.

References

Read only what the current move needs:

Read When
../leaf-soul/SKILL.md always: conduct, language, report shape, fact/guess separation, review handoff
references/gate-01-intent.md ① Intent pass/fail contract before locking intent
references/gate-02-unknowns-context.md ② Unknowns & Context pass/fail contract before closing Learn
../leaf-work/references/gate-authoring.md drafting, grilling, or revising durable Learn artifacts
../leaf-work/references/clarity-ledger.md choosing the weakest row to aim the next Learn question
../leaf-work/references/experiment-log.md a ② unknown needs an independent probe
../leaf-work/references/layout.md file layout, naming, file-vs-folder-by-count

Workflow

  1. Triage before Learn. Decide whether the idea should kill, defer, enrich, split, or run Learn. Capture is cheap; Learn starts only if the idea earns it.
  2. Create or resume the sprout. Use leaf init if needed, then leaf new <slug> unless a matching sprout already exists.
  3. Capture the snapshot. In 01-Learn/01-intent.md, preserve raw wording and current hunch. In 01-Learn/02-unknowns.md, record checked context and open questions. Also update 00-status.md ## Overview so a reader can see what this LEAF is doing without opening every gate file.
  4. Run ① Intent. Use references/gate-01-intent.md as the contract. Separate raw wording, sharp why, provisional what, and locked intent. Surface guessed facts before locking the intent. When ① changes the why, what, core noun, split decision, or output form, update the status overview in the same turn.
  5. Run ② Unknowns & Context. Use references/gate-02-unknowns-context.md as the contract. Group unknowns, gather references or internal facts, expose premises, and keep fact/assumption/user-review boundaries visible. When ② changes the viable direction, assumptions, scope, or next choice, update the status overview in the same turn.
  6. Ask the Learn-close question with the evidence in view. Before asking, show what was actually gathered so the user judges sufficiency from the files, not from memory. If anything remains, keep the sprout enriched and continue Learn. If the user confirms no Learn-blocking unknowns remain, recommend continue to Example.

Show the gathered references as a file tree first:

find .leaf/01-sprouts/<slug>/01-Learn/02-references -type f | sort

Render the result as a tree with a one-line note per file saying what it covers. An empty or thin tree is evidence too — name it plainly instead of hiding it; "no references were needed because " must be said, not implied.

Learn-close question:

Have you really learned and understood all the facts you need to know in order to judge and standardize this matter? Are there still facts to verify, cases to refer to, internal context, or unverified assumptions remaining?

Status Overview

00-status.md is the reader's table of contents for the LEAF. It is not the source of truth for detailed reasoning, but it must summarize the current shape:

  • request: the user's request in the user's words;
  • purpose: why this LEAF exists, after ① separates why from what;
  • expected output: the artifact, decision, model, document, code change, or result this LEAF is currently aiming at;
  • current scope: what is included, excluded, split, or still undecided;
  • consistency rule: reminder that this overview changes when the LEAF changes.

Do not let 00-status.md become stale. Whenever 01-intent.md, 02-unknowns.md, split decisions, Learn-close status, or a later return changes what the LEAF is doing, revise the overview before reporting back.

Learn Handoff

① produces the why, provisional what, and first questions ② must answer before the right what can be chosen.

② produces the facts, assumptions, candidate judgments, premises, and explicit review items that ③ Criteria consumes.

Do not continue to ③ merely because files exist. Continue only when ① and ② pass their gate contracts and the user has answered the Learn-close question.

Split Check

Run this before creating a sprout, when the idea branches, when the user adds a new direction, and before recommending ③.

Verdict Use when
split now bundled parts have independent core nouns, artifacts, success checks, reviewers, lifecycles, likely-change axes, or review/continuation paths
keep grouped parts are sequential concerns inside one outcome: one noun, one artifact, one acceptance check, one lifecycle
ask first splitting would decide the user's intent: the noun drifts, output form is exploratory, or a quieter sibling is not concrete enough

If split is clear and the user asked to capture the work, create or resume sibling sprouts. Otherwise recommend the split and name candidates. Do not carry a known mixed sprout into Example as one leaf unless the grouping reason is explicit.

Status Labels

Use these in 00-status.md:

Label Meaning
captured raw idea saved with minimal context
enriched meaningful context, references, premises, or alternatives were added
ready-for-example ① and ② passed; ③ Criteria can start in leaf-work in the same sprout
deferred parked until a named condition changes
killed not worth pursuing now

Do not mark active sprouts as fallen by editing status alone. fallen is a stage reached through an explicit leaf CLI action with a fallen reason.

Boundaries

  • Work only in 00-status.md and 01-Learn/ from this skill.
  • Do not fill ③ Criteria, ④ Wireframe, ⑤ Design, or tasks here.
  • If the user wants to build the selected artifact, draft, task graph, or execution path now, switch to leaf-work.
  • Reference and benchmark exploration belongs in ②; building from those references belongs in later gates.

Response Shape

Report per leaf-soul: overview first, decision points up top, facts separate from assumptions, and user-facing prose in the user's language.

Include briefly:

  • sprout path and status label
  • evidence checked
  • what was captured or changed
  • split/group/ask-first reasoning when relevant
  • recommendation: kill, defer, enrich, split, or continue to Example
  • next action if the user resumes later