تشغيل أي مهارة في Manus بنقرة واحدة

vast-transform

النجوم٢

التفرعات٠

آخر تحديث٢٤ مايو ٢٠٢٦ في ٠٦:١٩

Transform an abstract or prose description of a project, system, or initiative (NOT yet in VAST shape) into a VAST-shaped draft — extracting the latent Vision (or flagging its absence), mapping the rest to Architecture/Strategy/Tactics, and recommending composition depth. Generative, not analytical: it produces a proposed VAST document, it does not score one. Use this WHENEVER someone has raw prose (a pitch, a project brief, a backlog narrative, a strategy memo) and wants it reshaped, restructured, or re-expressed in VAST terms — EVEN IF they never say "VAST" by name, e.g. "put this in V/A/S/T terms", "what is the Vision/Architecture here", "structure this the way the framework would". Reshaping prose into clean layers (and honestly flagging a missing Vision instead of inventing one) is what this skill does better than freehand rewriting, so reach for it. Triggers: "transform this into VAST", "make this VAST-shaped", "extract the VAST from this", "VAST-ify this", "переведи в VAST". Not for scoring a doc alrea

التثبيت

التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.

تشغيل في Manus

المصدر

dkushnikov

dkushnikov/vast-framework

فتح مستودع GitHub عرض مستودعات المنشئ

تنزيل

تشغيل في Manus

المهن ذات الصلةSOC

استنادا إلى تصنيف SOC المهني

متخصصو إدارة المشاريعمهن الأعمال والعمليات المالية·SOC 13-1082

مستكشف الملفات

2 ملفات

SKILL.md

readonly

المزيد من هذا المستودع

نفس المستودع

vast-validate

dkushnikov/vast-framework

Validate a document that claims VAST shape — a single layer (e.g. a Vision doc) or a full V/A/S/T doc — checking that each layer is what it claims to be, scope is consistent across layers, and framework discipline holds. Detects layer drift (Vision written as a feature/deadline list, Architecture written as a project plan), mixed scope (company-level Vision + product-UI Architecture), missing Vision falsification triggers, incomplete composition frameworks, and Challenge Flow violations. Use this WHENEVER someone shares a Vision, Architecture, Strategy, or Tactics doc and wants it checked, reviewed, sanity-checked, or pressure-tested for whether the layers are clean and well-formed — EVEN IF they never say "VAST" or "layer purity". Checking layer discipline by hand is exactly what this skill does better: it applies fixed structural checks and returns a per-check Pass/Warn/Fail report, so reach for it rather than eyeballing. Triggers: "validate this VAST doc", "is this layer pure", "check VAST shape", "проверь

2026-05-242

vast-connect

dkushnikov/vast-framework

Verify that a child VAST doc genuinely cascades from its parent — that a product VAST nested under a company VAST (or a component VAST under a product VAST) is actually VAST-connected, not two unrelated docs wearing the same vocabulary. Checks scope step-down (child one recursion level narrower than parent), Vision lineage (child Vision refines / serves a parent Strategy use-case or Architecture domain), invariant inheritance (the core Matryoshka condition — child Architecture operates WITHIN parent invariants, never violates one), cross-doc Challenge Flow (child informs parent upward, parent constrains child downward — no inversions), and Strategy alignment (child invests with, not against, the parent direction). Use this WHENEVER someone has TWO VAST docs in a parent/child or nested relationship and wants to know whether they actually fit together, cascade, or cohere — EVEN IF they never say "VAST" or "Matryoshka". Tracing whether a child honors a parent invariant is exactly what this skill does better than

2026-05-242

vast-draft

dkushnikov/vast-framework

Scaffold a NEW VAST document from a minimal seed — a scope plus a sentence or two about an initiative — guiding the author through the four layers in commitment-priority order (Vision → Architecture → Strategy → Tactics). Generative, like vast-transform, but the input is near-empty rather than rich prose: the output is a SCAFFOLD (guiding questions + what-good-looks-like + fillable stubs), not a filled doc. Where the seed is silent it provides the question to answer, never an invented answer. Prompts for Vision falsification triggers and recommends composition depth. Use this WHENEVER someone wants to START a VAST doc and has only an idea plus a scope — EVEN IF they never say "VAST", e.g. "help me lay out a V/A/S/T for this", "give me the structure to fill in". Producing a commitment-ordered scaffold (and refusing to fabricate answers the seed did not give) is what this skill does better than a blank template, so reach for it. Triggers: "draft a VAST", "scaffold a VAST doc", "start a VAST for", "new VAST from

2026-05-242

vast-okr-audit

dkushnikov/vast-framework

Audit an EXISTING set of OKRs (objectives and key results that are already written) for VAST-triad conflation — objectives that are really Outputs (a project or feature like 'ship the new billing system' dressed up as an outcome), key results that fuse two layers (a strategy-level Use Case and a measurable Output crammed into one KR), and scope inconsistency (company-level and team-level objectives in the same set). Use this skill WHENEVER someone hands you OKRs — freshly drafted, in-progress, or finalized — and wants them checked, sanity-checked, sense-checked, reviewed for structure, or asks whether the objectives are genuine outcomes versus relabeled projects, whether the set is 'clean', or whether the objectives and key results sit at the right level — EVEN IF they never say 'VAST', 'conflation', or 'audit'. Reviewing OKR structure by hand is exactly what this skill does better than ad-hoc reading: it applies fixed structural heuristics and returns a per-OKR Pass/Warn/Fail report, so reach for it rather t

2026-05-242

name

vast-transform

description

vast-transform

When to use

Invoke when you have an abstract or prose description — a pitch, a project brief, a backlog narrative, a strategy memo, a feature list — that is not yet in VAST shape, and you want it reshaped into a proposed Vision / Architecture / Strategy / Tactics draft. This skill is generative: it produces a VAST-shaped document from prose, extracting the latent Vision and sorting the rest into layers.

Common triggers:

"Transform this into VAST"
"Make this VAST-shaped" / "VAST-ify this"
"What's the Vision / Architecture here?"
"Put this in VAST terms"
"Переведи это в VAST"

This skill is distinct from the analytical siblings. vast-validate checks a doc that already claims VAST shape and emits Pass/Warn/Fail. vast-okr-audit audits OKR-shaped docs for triad conflation. vast-transform takes prose that makes no VAST claim and produces the shape. If the input already has V/A/S/T headings and you want it scored, use vast-validate instead.

VAST recap (condensed)

The output sorts prose into four layers, each owning a distinct level of commitment:

Layer	Owns — what prose content maps here
Vision	The why / for-whom / enabled-experience. The purpose the work serves, as a falsifiable hypothesis. Look for "we believe", "so that", "for {users} who {pain}".
Architecture	The structural domains and the composition framework — skill library, interfaces, invariants, implementations. The durable structure work is built on. Look for "the system is structured as", "capabilities", "guarantees", "what must always hold".
Strategy	Investment direction and sequencing — which experiences to compose next, in what order, validated how. Look for "first we'll", "prioritize", "sequence", "invest in", "validate with".
Tactics	Concrete deliverables — specific compositions for specific users at specific moments. Look for named features, screens, dated artifacts, per-user delivery.

Two cross-cutting facts the transform respects: Vision is a committed falsifiable hypothesis (it needs named revision triggers — flag their absence as a gap, don't fabricate them), and the four layers form a commitment-priority order (Vision → Architecture → Strategy → Tactics), not a temporal sequence.

For the four layers, the triad (Outcomes / Use Cases / Outputs), and the invariants/implementations split, see ../../references/vast-essentials.md. For the four-scope model (company / product / function / WoW) and per-scope identifying signals, see ../../references/layer-definitions.md. The eight common confusions — which the produced output must not exhibit — are in ../../references/anti-patterns.md. The composition-depth spectrum (deep / medium / light / vocabulary-only) is in the framework's applicability.md. Detailed transform heuristics are in references/transform-heuristics.md — read those if any extraction or mapping is ambiguous.

Process

For the prose under transformation:

Identify scope — company / product / function / WoW. State it explicitly; it determines what each layer should contain and what "good" looks like. (Scope signals: references/transform-heuristics.md#scope-identification; the full per-scope signal lists are in ../../references/layer-definitions.md.)
Extract Vision — find the why / for-whom / enabled-experience in the prose. If it is genuinely present, lift and tighten it. If it is genuinely absent (the prose is all features/deliverables with no stated purpose), flag the absence — do not fabricate a confident Vision. Offer a candidate Vision clearly labeled ⚠️ CANDIDATE (confirm) that the user must confirm. (See references/transform-heuristics.md#vision-extraction.)
Map to Architecture / Strategy / Tactics — sort the remaining prose into layers: structural domains / composition framework → Architecture; investment direction / sequencing → Strategy; concrete deliverables → Tactics. Note thin or empty layers rather than padding them. (See references/transform-heuristics.md#layer-mapping.)
Recommend composition depth — deep / medium / light / vocabulary-only. Match depth to the work's actual AI-substrate engagement and scope; do not force deep where light fits (that is anti-pattern #6) nor claim vocabulary-only where deep is needed (anti-pattern #7). (See references/transform-heuristics.md#depth-recommendation.)
Surface gaps — what's missing for a complete VAST doc: no falsification triggers on the Vision, Architecture with no invariants, no for-whom in the Vision, no named accountable owner, scope ambiguity. (See references/transform-heuristics.md#gap-surfacing.)
Emit the proposed VAST doc per Output format below.

Heuristics

#	Heuristic	One-liner	Detail reference
1	Scope identification	Match prose against per-scope signals; pick ONE scope (company / product / function / WoW). If the prose mixes altitudes (company ambition + product-feature detail), pick the dominant scope and record the conflict as an open question — don't silently average.	`references/transform-heuristics.md#scope-identification` (primary); `../../references/layer-definitions.md` (per-scope signals)
2	Vision extraction	Lift why/for-whom/enabled-experience if present. If absent, mark status ABSENT and offer a `⚠️ CANDIDATE (confirm)` Vision derived from the deliverables — never present an inferred Vision as settled fact. Partial why → "partially inferred".	`references/transform-heuristics.md#vision-extraction` (primary); `../../references/anti-patterns.md` (#1 Vision-as-Use-Case, #4 missing falsification triggers)
3	Layer mapping	Sort prose: structural domains / composition framework → Architecture; investment direction / sequencing → Strategy; concrete deliverables → Tactics. Don't pad a thin layer — note what's needed. Don't promote a feature to Architecture just to fill it.	`references/transform-heuristics.md#layer-mapping` (primary); `../../references/vast-essentials.md` (triad; invariants/implementations)
4	Depth recommendation	Pick deep / medium / light / vocabulary-only by scope × AI-substrate engagement. One-line reason citing the signal that decided it. Don't force deep on light work (#6) or claim vocabulary-only where deep is needed (#7).	`references/transform-heuristics.md#depth-recommendation` (primary); `../../references/anti-patterns.md` (#6, #7)
5	Gap surfacing	Enumerate what's missing for a complete VAST doc: Vision falsification triggers, Architecture invariants, for-whom, named accountable owner, scope clarity. Gaps are honest absences to report, not things to invent.	`references/transform-heuristics.md#gap-surfacing` (primary); `../../references/anti-patterns.md` (#2 Architecture-by-default, #7 vocabulary-only)

Heuristic numbers above are stable identifiers, not an execution order. The Process runs scope → Vision → mapping → depth → gaps. Reference heuristics by name (e.g. "Vision extraction") if you annotate the output.

Output format

Produce a markdown document with this exact structure — a proposed VAST doc, not a score:

# Proposed VAST — {title} ({scope} scope)

## Vision
{extracted Vision — why / for-whom / enabled-experience. If genuinely absent,
write a single clearly-labeled candidate instead:
"⚠️ CANDIDATE (confirm): {inferred purpose} — derived from the deliverables below;
confirm or replace before use."}

## Architecture
{mapped structural content — domains, composition framework, invariants.
If thin or empty, state what the prose provides and what a complete Architecture
still needs (e.g. "no invariants stated — needs ≥1 quality + ≥1 fallback contract").}

## Strategy
{mapped investment direction / sequencing. If absent, note that no sequencing
logic was stated.}

## Tactics
{mapped concrete deliverables — features, screens, per-user delivery.}

---

## Notes
- **Vision status:** extracted | partially inferred | ABSENT (candidate offered — confirm before use)
- **Composition depth:** deep | medium | light | vocabulary-only — {one-line reason}
- **Gaps surfaced:** {bullets — what's missing for a complete VAST doc}
- **Open questions:** {bullets — what the author must decide}
- **Decisions deferred:** {bullets, if any}

Always include all five Notes fields. Where a field is empty, write (none) — keep every field present so the Notes block is identical in shape across clean and gap-heavy transforms (mirrors the exemplars' empty-state discipline).

The four layer sections always appear, even when a layer is thin or empty — an empty layer is itself a finding (state what's missing), not a section to omit.

Examples

See:

../../examples/07-transform-clean/ — prose with a clear latent Vision + identifiable A/S/T material → clean transform, Vision status "extracted".
../../examples/08-transform-no-vision/ — a pure feature/deliverable list with no stated why → A/S/T mapped, Vision status "ABSENT", labeled candidate offered (not fabricated).
../../examples/09-transform-mixed-altitude/ — prose mixing company-level ambition with product-feature detail → transform picks a scope, records the altitude conflict as an open question.