| name | faffter-noon-intake |
| description | Default `intake` / discovery producer — turns a loose idea, bug, or empty repo into a structured discovery brief for faff-jot/faff-plot to shape into tickets. The light conversational default. Runs as a configured slot, not the user `/` menu. |
| user-invocable | false |
| judgement_seam | modedetect |
faffter-noon-intake
The default intake / discovery producer. Turns a loose starting point — "I want to build X", a bug report, a half-formed feature idea, or an empty repo — into a structured discovery brief that the orchestrator hands to the methodology slot for shaping into tickets.
This is the implicit default for the intake slot when none is configured. It is the light, conversational counterpart to heavier ideation skills (superpowers:brainstorming, gstack:office-hours): same slot, same output contract, less ceremony. Extracted here so it can be invoked standalone, tested, and swapped.
slots:
intake: faffter-noon-intake
When it runs
Invoked by /faff-jot as the configured intake skill (the default when the slot is unset). /faff-jot decides the mode and passes it in; this skill runs the discovery conversation and returns the brief. It does not create tickets, touch the tracker, or write code — that is the orchestrator's job (/faff-jot), which applies the methodology slot to the brief this skill produces.
Can also be invoked standalone: "help me scope a new feature" produces a discovery brief for review without the tracker lifecycle.
Input
The caller (/faff-jot) provides:
- Mode —
greenfield (kick off an empty/new project) or single-item (a new feature, bug, or ticket inside an existing project).
- The human's starting description (whatever they said — a sentence, a paragraph, a pasted bug report).
- For
single-item: context on the existing project — the live workstream/container list, naming conventions, and any current-priority signal (from the tracker + the consuming project's CLAUDE.md), so discovery can place the item correctly.
- For
greenfield: any existing repo signal (language, framework, README) if the directory isn't truly empty.
How it runs — discovery, not interrogation
A short, focused conversation. The goal is to extract just enough to shape tickets, not to write a spec (that's /faff-prep's job, later, per ticket). Pull on the threads that change how the work breaks down; leave implementation detail for prep.
greenfield mode
Establish the shape of the whole thing:
- What is it / who is it for — the product or system in one or two sentences, and the user it serves.
- The major capabilities — the handful of distinct things it must do. These become candidate workstreams.
- The first slice — what "v0" or "the first useful version" looks like. This is what sequences to the front.
- Hard constraints — language, framework, platform, deadline, anything non-negotiable that governs how work is grouped.
- Explicit non-goals — what this is deliberately not, so the methodology doesn't manufacture tickets for it.
single-item mode
Establish enough to write one well-formed ticket (or a small set if it splits):
- Intent — what the user wants to happen that doesn't today (feature), or what's broken (bug).
- Done signal — how you'd know it works. Rough acceptance signals, not full criteria.
- Affected area — which part of the existing system / which workstream this belongs to.
- Known dependencies — anything that must ship first, or that this enables.
- Size sense — is this one ticket, or does it obviously span two independent concerns? (Don't force a split — just flag it for the methodology to decide.)
Stop pulling threads once the brief below can be filled. Unresolved questions are captured, not chased — they go into the brief as open questions for prep to resolve, not as blockers here.
Output — the discovery brief (the intake slot contract)
A single markdown document in the shape below. This is the intake-slot output contract: the orchestrator and the methodology slot's ticket-shaping output both consume it, so any skill placed in the intake slot (including superpowers:brainstorming or gstack:office-hours) must emit this shape. Producers may add sections; they must not omit the required ones.
# Discovery brief: <short name>
mode: greenfield | single-item
## Goal
<1–3 sentences: what we're building / fixing and who it's for.>
## Capabilities # greenfield: the major distinct areas (workstream candidates)
- <capability> — <one line on what it covers>
...
## Item # single-item: the one feature/bug, replaces Capabilities
intent: <what should happen / what's broken>
done-signal: <how you'd know it works>
area: <which existing workstream / part of the system>
split-sense: unitary | possibly-splittable (<the two concerns, if so>)
## First slice # greenfield only: what v0 is — sequences to the front
<the smallest useful version>
## Constraints
- <hard constraint: language / framework / platform / deadline / convention>
...
## Dependencies
- <known prerequisite or enabled-by relationship, named as concretely as possible>
...
## Out of scope
- <explicit non-goal> — <one line>
...
## Open questions
- <thing the human didn't resolve — carried forward for /faff-prep, not blocking>
...
Rules for the brief:
- Required sections:
Goal, exactly one of Capabilities (greenfield) or Item (single-item), Constraints, Out of scope. The rest are included when they have content; an empty section is omitted, not left blank.
- No tickets here. The brief describes the shape of the work; turning it into tickets (titles, relationships, sequencing, container placement) is the
methodology slot's ticket-shaping output, applied by the orchestrator. Do not pre-write ticket titles or pick a structure — that is a methodology decision, and hard-coding it here defeats the point of the swappable methodology slot.
- Names over invented codes. Capabilities and dependencies are named in plain English, not labelled
C1/D2. (Same skimmable writing-style rules the rest of faff follows.)
- Greenfield ≠ a spec. The brief is the input to ticket creation, not a build spec. Per-ticket specs are produced later by
/faff-prep.
Appetite
This skill reads the suite-wide appetite setting but has little to modulate — discovery is inherently a conversation with the human, so it runs the same at every level. The one effect: at low/medium appetite it asks more clarifying questions before closing the brief; at high/full it makes reasonable assumptions on minor ambiguities and records them under Open questions rather than stopping to ask. It never invents goals the human didn't express.
Rules
- Discovery extracts what changes how work breaks down — no more. Resist drifting into implementation detail; that belongs to
/faff-prep.
- This skill produces the brief; it does not create tickets, mutate the tracker, or write code. The orchestrator (
/faff-jot) owns those, applying the methodology slot to the brief.
- The discovery-brief shape is the slot contract. A third-party
intake skill conforms by emitting the same required sections; the orchestrator then shapes and creates tickets identically regardless of which intake skill ran.
- Unresolved questions are captured in the brief, never used as a reason to refuse to produce one. A thin brief with open questions is valid output — prep resolves the rest.