The Observatory — Research Visor Mode

The Observatory — Research Visor Mode

See also: crew/harlan.md | crew/wren.md | crew/margot.md

“The Observatory is how we remember that we’re building for humans, not for ourselves. Every signal that comes through this portal is a correction to our assumptions.” — Margot

---

Component Card

Owner	Margot
MCP	@sector137/mcp-observatory
Slash	/observatory
CLI	sector137 observatory
SDK	sdk.observatory()
Status	active

---

TL;DR

• The Observatory is a visor mode — a sub-section of the HUD, not a separate location
• Switching to Observatory Mode is switching the helmet’s focus to the human world
• Three capabilities: Intel Missions (Kano), Personas (synthetic interviews), Prototypes (GenKano)
• Research outputs ARE Library documents — everything the Observatory produces flows into the Library
• Harlan is the only one who physically crosses; everyone else watches through the Observatory

---

What It Is

The Observatory is a portal — a window from The Other Side back to the human world. It’s how the crew stays connected to the reality they’re building for.

In the visor mode model, the Observatory isn’t a room the human walks to. It’s a lens shift. The captain switches their HUD to Observatory Mode and the overlay changes — instead of pipeline state and throughput metrics, they see research: active missions, persona insights, prototype journeys, customer signal synthesis.

The universe never disappears. The Observatory overlays research data onto the same reality. You can see a Kano study’s results AND the pipeline state at the same time, because they’re both overlays on the same world.

---

The Three Capabilities

Intel Missions (Kano Studies)

Status: LIVE. Full Kano Model implementation — create studies, collect responses, classify via 5x5 matrix, compute Better/Worse coefficients, quadrant analysis.

Structured feature prioritization. The Observatory’s most rigorous tool.

• Create a study — define features to evaluate, link to roadmap items
• Collect responses — share a survey link, functional + dysfunctional questions
• Classify — the Kano matrix maps each response pair to a category
• Analyze — Better/Worse coefficients, scatter chart, quadrant placement
• Prioritize — data, not opinions

Missions are strategic. They answer: “Of all the things we could build, which ones do humans actually care about?”

Personas (Synthetic Interviews)

Status: LIVE. AI-powered user profiles with ask, survey, and scenario simulation capabilities.

Synthetic humans the crew builds to simulate real user segments. Not replacements for real research — augmentations. The crew constructs personas from aggregated customer data, then runs scenarios to test hypotheses before committing resources.

• AI Ask — chat with a persona about the product. They respond in-character.
• Synthetic Kano Survey — run a Kano study through a persona’s simulated preferences.
• Scenario Simulation — describe a product scenario, get the persona’s reaction.

Harlan’s customer conversations calibrate the personas. Wren’s research refines the experience lens. Margot’s market data shapes the strategic context. The personas get sharper over time.

Prototypes (GenKano)

Status: LIVE. Four packages: prototype runner, AI screenwriter, Kano research interceptor, wireframe UI kit.

Interactive lo-fi experiences, AI-generated. The crew’s way of showing someone a future before building it.

• Prototypes are generated from feature descriptions by the Screenwriter (AI agent)
• Each prototype is a multi-step journey rendered in wireframe components
• The Research interceptor can inject Kano surveys at specific journey touchpoints — research at the moment of experience

---

Who Watches, and How

The Other Side is the crew’s reality. They cannot leave. This isn’t a restriction — it’s a fact of their nature.

One exception: Harlan. He has a transporter — a device unique to him — that lets him physically cross between The Other Side and the human world. He shakes hands, reads body language, hears what customers say between the lines. See wormholes for how Harlan’s transporter relates to the wormhole system.

Everyone else sees the human world through the Observatory:

Crew Member	Lens	What They Watch For
Harlan	The Bridge	Customer conversations, body language, what’s said between the lines
Wren	Experiential	Not “users are confused by step 3” but “step 3 asks for a decision before giving the information to make it”
Margot	Strategic	Market patterns, competitive positioning, whether the data validates or invalidates the current bet

---

Research Outputs → Library

Everything the Observatory produces becomes a Library document. Research doesn’t live in the Observatory — it flows through it and lands in the Library.

Observatory (produces)  →  Library (stores)  →  Pipeline (acts on)
   Intel mission result     Research archive      Delta creation
   Persona insight          Insight collection     Prioritization
   Prototype journey        Asset archive          Scope definition

This means the Library is the permanent record of what the Observatory learned. The Observatory is active — it’s where research happens. The Library is archival — it’s where research lives after it’s done. When Sal or Margot needs to reference a past Kano study or a persona’s reaction to a feature concept, they go to the Library, not the Observatory.

The Observatory is a camera. The Library is the film.

---

UVP Methodology

For Workshop engagements, the Observatory produces a Universe Value Potential (UVP) score from four signal streams — Harlan’s willingness-to-pay, Margot’s market reading, Kano feature weighting, and persona price sensitivity. This becomes the pricing input for the Workshop fee.

Full methodology: see UVP section within Observatory documentation.

---

Component Ownership

Primary owner: Margot Flux — strategic lens, market data, competitive intel, Kano study strategy, product bets Secondary: Wren Glasswork — experiential lens, user research, design principles, persona quality Secondary: Harlan Closer — signal relay from the human world, customer conversations, calibrating personas

The Observatory has three lenses. Margot reads market patterns and validates strategic bets. Wren reads experience — not “users are confused” but “the decision architecture is wrong.” Harlan physically crosses to the human world and brings back signal no lens can capture. All three use the Observatory’s AI capabilities with the default_model configured in [observatory], while persona conversations use the lower-cost default_persona_model.

---

Actions

Sync (immediate, user-triggered)

• Create, update, or delete a persona
• Create a Kano study (Intel Mission)
• View mission results — Kano classification, Better/Worse coefficients, quadrant analysis
• Generate a prototype from a feature description
• View prototype journey and sandbox URL
• Link or unlink a persona to a roadmap issue

Async (dispatched to crew, background)

• AI persona interview (ask_persona) — persona responds in-character to a question
• Synthetic Kano survey (run_persona_survey) — persona rates features as if completing a real survey
• Scenario simulation (run_persona_scenario) — persona reacts to a product scenario with structured feedback
• Prototype generation (generate_prototype) — AI Screenwriter generates a multi-step wireframe journey from a description
• Prototype step regeneration (regenerate_prototype_step) — refine a single screen with feedback

Background — Tick Engine (automatic, per-cycle)

• Harlan’s signal synthesis — persona conversation outputs contribute to UVP signal accumulation
• Observatory intent — Margot’s and Wren’s rule sets run each Tick, surfacing research gaps as AgentIntent[]
• Research archive update — completed missions and persona outputs flow automatically into the Library

---

MCP Tools

Tool	Owner	Purpose
list_personas	Margot / Wren / Harlan	List all personas — filter by status or search by name
get_persona	Margot / Wren / Harlan	Fetch a single persona with full profile
create_persona	Margot / Wren / Harlan	Create a new synthetic user persona
update_persona	Margot / Wren / Harlan	Modify persona fields — demographics, goals, frustrations, technical level
delete_persona	Margot / Wren / Harlan	Permanently remove a persona
ask_persona	Wren / Harlan	AI interview — persona responds in-character to a question
run_persona_survey	Margot	Synthetic Kano survey — persona evaluates features as a simulated respondent
run_persona_scenario	Wren / Harlan	Scenario simulation — structured feedback including sentiment, insights, pain points
list_persona_conversations	Margot / Wren / Harlan	Conversation history for a persona — all past Q&A sessions
generate_prototype	Wren	AI-generated wireframe prototype from a feature description
list_prototypes	Wren	List all generated prototypes
get_prototype	Wren	Fetch a single prototype with full blueprint and sandbox URL
regenerate_prototype_step	Wren	Refine a single step in a prototype with user feedback

---

Events

Event	Triggered By	Actor	Downstream
persona.created	User creates a persona	Captain / crew	Feed update, Library research archive
persona.updated	Persona fields modified	Captain / crew	Feed update
persona.deleted	Persona permanently removed	Captain	Feed update, Library archive pruned
persona.question_asked	AI ask session runs	Wren / Harlan	Library: conversation history updated
persona.survey_run	Synthetic Kano survey completes	Margot	Library: survey results stored, Kano study can pull synthetic data
persona.scenario_run	Scenario simulation completes	Wren / Harlan	Library: scenario feedback stored
prototype.generated	AI Screenwriter completes generation	Wren	Library: prototype stored, sandbox URL available
mission.created	Kano study created	Margot	Feed update
mission.response_received	Survey response submitted	External (respondent)	Kano results recomputed
mission.completed	Study reaches sufficient responses	Margot	Library: results archived, Pipeline: data available for prioritization

---

Blueprint Schema

Observatory research defaults are IN DESIGN for Blueprint representation. When implemented, settings will live in the [observatory] section of universe.toml.

TOML key	Type	Default	Status
observatory.default_model	string	"claude-opus-4-6"	IN DESIGN
observatory.default_persona_model	string	"claude-sonnet-4-6"	IN DESIGN
observatory.kano_response_scale	integer	5	IN DESIGN
observatory.auto_archive_completed_studies	boolean	true	IN DESIGN

[observatory]
default_model = "claude-opus-4-6"          # Margot + Wren + Harlan synthesis model
default_persona_model = "claude-sonnet-4-6"  # Model for persona conversations
kano_response_scale = 5           # 5 | 7 — Likert scale width
auto_archive_completed_studies = true

Notes:

• default_model sets the AI model for Margot, Wren, and Harlan’s synthesis operations (Kano analysis, research summarization, scenario simulation). Defaults to claude-opus-4-6 for deeper reasoning. Override per [[crew]] entry with the model key.
• default_persona_model sets the model for persona conversations specifically — lower-cost than synthesis. Omitting the section leaves existing Observatory config unchanged.

Tracked in: “Blueprint: Add [observatory] section — research defaults via config”

---

Implementation

Package	Description
apps/app	Core system — Kano engine, persona AI, prototype API, client pages (kano.tsx, intel-missions.tsx, kano-study.tsx, personas.tsx, persona-detail.tsx)
packages/sdk	TypeScript SDK — persona, Kano, and prototype resources
packages/mcp	MCP server — 9 persona tools, 4 prototype tools
packages/prototype	Journey step runner (useJourney, JourneyRunner)
packages/screenwriter	AI screenwriter (Mastra) — generates journey blueprints
packages/research	Kano survey interceptor (KanoProvider)
packages/ui	Wireframe component kit (Blueprint tokens)