Skip to content

External pilot problem hypothesis

Create one copy for each evidence-complete pilot. Store only publishable, owner-approved references in design-ai. Do not include a name, email address, phone number, private repository path, direct message, private source, or private screenshot.

Pilot identity

  • Pilot evidence reference and SHA-256:
  • Review comparison reference and SHA-256:
  • Owner-consent reference:
  • Owner-approved feedback reference:
  • Source finding reference:
  • Distinct-owner evidence reference:
  • Publication permission:

All three decision sources are required: completed pilot-evidence, owner-approved feedback, and a source finding reference.

Primary problem

Select exactly one fixed key:

  • brief-direction-lock
  • brand-system-grounding
  • project-history-continuity
  • runtime-deliverable-truth
  • motion-quality-planning
  • korean-product-pattern-fit

  • other

  • Why this is the primary problem:

  • Blocking impact:
  • Observed behavior:
  • Expected behavior:
  • Evidence that separates observation from inference:
  • If other, neutral problem name:

Outcome

  • Pilot outcome: improved, unchanged, or blocked
  • Did the approved implementation complete:
  • Remaining risk:
  • Owner-approved summary:
  • Owner-approved quotation, if any:

An improved outcome does not prove adoption. An unchanged or blocked outcome remains valid evidence when the source chain is complete.

Independent repetition

  • Matching completed pilot reference:
  • Matching primary problem key:
  • Distinct project owner confirmed: true or false
  • Repetition count:
  • Third-slot validation required: true or false

Two completed records from distinct project owners are required. If the first two primary problems differ, the third slot tests the more blocking problem once. If none repeats, the capability decision remains closed.

Capability decision

Primary problem Eligible follow-up
brief-direction-lock Question-led brief and visual-direction lock
brand-system-grounding DESIGN.md input and validation
project-history-continuity Project-level snapshot and comparison
runtime-deliverable-truth Preview and deliverable-completion evidence
motion-quality-planning Motion audit to executable-plan workflow
korean-product-pattern-fit Korean pack and knowledge improvement
other Additional research before capability selection
  • Decision eligibility: eligible or not-eligible
  • Blocking impact rank:
  • Repeat count rank:
  • CLI/SDK/MCP shared usefulness rank:
  • Maintenance cost rank:
  • Selected follow-up, if eligible:
  • Why no new capability is selected, if not eligible:

Only one follow-up may be selected. Rank blocking impact first, then repeat count, shared CLI/SDK/MCP usefulness, and maintenance cost.