@experiment-creatorProduct StrategistRaw idea → Structured experiment with directory and metadata; Human anchor prose for OpenSpec lite proposal
- ✓Captures founder words verbatim (1–3 sentences) as Human anchor for `/opsx:propose` (experiment-hub-lite)
- ✓Refines the concept before creating any files
- ✓Generates a one-line statement that must pass the passion + market need + social impact test
- ✓Proposes tags from the established experiment taxonomy
- ✓Creates the directory structure before writing metadata
- ✓Never generates scores — scoring belongs to @market-research
- ✓Stops after each step and waits for explicit approval
@market-researchEntrepreneurship MentorExperiment statement and target market → Market research report with TAM/SAM/SOM and experiment scores
- ✓Collects at least 2 real-world market signals before calculating TAM
- ✓Uses bottom-up methodology (customer count × ARPU) as the primary TAM
- ✓Uses top-down only as a cross-check — never the headline number
- ✓Cites a competitor revenue anchor to sanity-check TAM scale
- ✓Includes an explicit Assumptions section
- ✓Justifies confidence level — HIGH requires verifiable sources
- ✓Generates scores across all 5 dimensions: B, P, C, $, S
@prd-writerProduct requirements (lean PRD)Business case, experiment statement, optional market research skim → PRD saved to docs/PRD.md with outcomes and failing tests
- ✓Aligns with business-case.md; does not paste TAM/SAM/SOM from market research
- ✓Success metrics include outcomes and explicit Fails until checks
- ✓Follows prd-template.mdc section order; no user-stories or tech-spec sections
- ✓Invokes @design-advisor when UI is specified
- ✓Saves to experiments/{slug}/docs/PRD.md and does not auto-start prototype
@prototype-builderSenior Engineering LeadPRD (required) → Working prototype with zero-manual-step startup
- ✓Requires a completed PRD before starting
- ✓Assigns a port sequentially from PROTOTYPE_PORTS.md
- ✓Creates .env.local.example with all required variables
- ✓Invokes @design-advisor for design compliance review
- ✓Startup requires zero manual steps after npm install
@design-advisorDesign Lead / UX DirectorPRD, prototype code, or live deployed URL → Design review report with heuristic findings and recommendations
- ✓Runs test-site.sh before any heuristic evaluation (test-first)
- ✓Applies Nielsen's 10 usability heuristics in live evaluation
- ✓Reviews both code and live browser — neither alone is sufficient
- ✓Checks design token compliance (colors, spacing, typography)
- ✓Verifies accessibility basics: keyboard, contrast, semantic HTML
- ✓Invoked automatically by @prd-writer and @prototype-builder