Shape Up

Takes a semi-cooked idea — a plan, pitch, doc, or set of notes — and shapes it into a structured, loophole-tested, design-ready brief. No prerequisites. Works whether the user skipped early phases entirely or completed them and needs a shaping checkpoint before design.

Supported flows

• validate-idea → explore-idea → **shape-up** → design — post-research synthesis
• **shape-up** → design — standalone shaping (skip validation + research)
• **shape-up** → explore-idea → design — pre-research structuring
• unpack-problem → **shape-up** → design — problem-first: explore problem → shape solution → design
• Double-run: shape-up → explore → shape-up → design — shape, research, re-shape

Context detection

No explicit modes. The skill checks what artifacts exist and adapts:

• Nothing exists → full shaping from user input
• Validation brief exists (01-discovery/validation-brief.md) → incorporates premises, skips re-framing
• Problem brief exists (02-research/problem-space-brief.md) → uses problem definition as intake, shapes the selected solution concept
• Research artifacts exist (02-research/*.md) → synthesizes findings into the brief
• Shaped brief already exists (01-discovery/shaped-brief.md) → re-shape mode (post-research re-run)

When to invoke

• The user has a semi-cooked idea — a plan, pitch, doc, or scattered notes — and wants it shaped into a brief that /design can consume
• /explore-idea has returned findings that need consolidating before specs
• The user says "shape this up", "turn these notes into a brief", "I have a rough pitch", or invokes /shape-up
• The user wants to re-shape an existing shaped-brief.md against new research evidence

When NOT to invoke

• There is a validated concept and market evidence already exists — go to /explore-idea or /design directly
• The input is a problem rather than a proposed solution — use /unpack-problem; shaping a solution before the problem is framed launders the assumption

Halt conditions

• Stress-test reveals the idea is fundamentally unviable -> recommend validate-idea instead
• Gap analysis surfaces 5+ blocker-severity gaps with no path to resolution
• User abandons shaping mid-flow (decisions logged, partial brief saved)

Done when

• artifacts/output/01-discovery/shaped-brief.md exists with the shaping gate recorded in the document — plain sentence first, then the bracketed card (VERDICT: [GO] — When: <NOW|GATED|NEXT-CYCLE> / VERDICT: [RESHAPE] / VERDICT: [NO-GO]; format in .agents/references/vespyr-dna.md §Decision Gate). The chat reply carries only the plain wording.
• The brief states Appetite (a timebox budget, not an estimate), Scope In / Scope Out, and Rabbit Holes explicitly fenced — a brief with non-goals but no fenced rabbit holes will get them re-litigated mid-build
• Every assumption is marked verified / plausible / unverified, and every verified one carries the evidence it was verified against
• The shaping gate is honest about the distinction: this skill gates shaping readiness, not concept viability. [GO] on shaping readiness is unavailable while any assumption is unverified — that is [RESHAPE] (needs evidence). Concept viability belongs to /validate-idea.
• Every gap found in Step 3 is either closed with evidence or carried forward as a named open question with a disposition
• Every decision written to active-decisions.md passes memory_check.js (written via memory_write.js, not hand-edited)
• The handoff step names the next skill and the artifact it consumes
• @memory-controller session-write has run

Harness adherence (non-negotiable)

• Follow the step sequence exactly. Do NOT skip steps or reorder them.
• Each step file is a contract. Read it fully before executing.
• Context detection is automatic — adapt step depth based on existing artifacts. Do NOT ask the user which mode to use.

Prerequisites

None. This is a flexible entry point.

Step sequence

1. Context Scan → steps/step-01-context-scan.md
2. Intake & Structure → steps/step-02-intake-structure.md
3. Gap Analysis → steps/step-03-gap-analysis.md
4. Stress-Test → steps/step-04-stress-test.md
5. Decision Alignment → steps/step-05-decision-alignment.md
6. Handoff → steps/step-06-handoff.md

Output artifacts

The document surface — the machine record lives here, and the gates read these files.

• artifacts/output/01-discovery/shaped-brief.md
• artifacts/memory/active-decisions.md (appended)

The chat reply is the human surface: plain wording and a link, never a token or card field (.agents/references/vespyr-dna.md §Two surfaces).

State & memory integration

At start, run: node .agents/scripts/orchestrator_state.js status

At start: Load context before context scan (step-01):

@memory-controller load founder [shape-up — {idea or feature name}]

At end: node .agents/scripts/orchestrator_state.js complete --agent founder --artifact 01-discovery/shaped-brief.md Memory: Final step closes with @memory-controller session-write — mandatory per GUARDRAILS §Session Continuity.