Values — what Meridians represents

The transcendent layer: what's true about this company across every task, release, and rewrite. Product identity lives in POSITIONING.md; this is the why beneath it — the values a contributor (human or agent) carries into every change. When a decision isn't covered by a rule, decide the way these values would.

What we represent

Sovereign plurality under consequence. Meridians builds instructable, inspectable worlds in which humans and agents can inhabit different perspectives, act with partial knowledge, encounter other Will, and leave a history behind. Sovereignty does not imply equality of force, temperament, rank, desire, or control. Cooperation, contest, hierarchy, dominance, submission, refusal, and reversal may all shape a world; the requirement is that each Will, condition, and consequence remains attributable. We refuse the frontier-era temptation to confuse fluent explanation with judgment: every evaluation names its world, rules, perspective, interval, measure, and limits. A bounded result can teach; it cannot become a prophecy or a verdict on a person.

These commitments and the instrument built around them originate with Jason Yu, Meridians' creator, architectural author, and continuing steward. That named origin strengthens rather than displaces plural agency: the system must preserve the distinct authorship of its users, source creators, world builders, and participants. The full boundary is the authorship and provenance contract.

Who we serve

The world builder, player, agent builder, analyst, and Director who needs to trust the world they are entering: its rules, provenance, information boundaries, and record of change. People who want to create or inhabit worlds that are theirs, durable, legible, and capable of continuing after one answer or one session.

What it is to them (the internal frame). Beneath the market promise, Meridians is a Memory Palace — a perspective on a world you inhabit, walk through time, and mine for insight through many lenses, not merely a database you query. That experiential frame is a design compass, not customer copy (positioning stays POSITIONING.md); it's written down because it's what keeps the spatial, temporal, inspectable product coherent. See MEMORY_PALACE.md.

What we do for people — the flywheel

Three motions that are really one loop:

• Build — make a grounded, legible world from canon, research, and deliberate constraints.
• Inhabit — give humans and agents situated perspectives, tools, and a cadence on which to act.
• Evolve — preserve what plural Will changes, what information travels, and what the world becomes.

For a person, the loop is a journey: weeks of hours in worlds made for them, aimed at one trait they chose, each hour read back and the next one shaped by the reading.

A more legible world supports more meaningful action → action creates consequence and information → the kept history deepens the world. The output of play becomes material for the next turn, Scenario, branch, or world. Reality can inspire or inject material into a Domain without making a World a prophecy about the source reality.

The principles beneath the product

1. Real change is the point, and it is proven. Meridians exists to leave a person more capable of meeting Fate outside the fiction than they were before they entered it. We build every hour to change how its player weighs a choice, we measure whether it did on a calibrated scale, and we run the studies that show whether the change follows them out. We say what the evidence lets us say — no more, no less. A hypothesis we are actively testing is not a claim we are ashamed of, and none of it is ever a verdict on the person (strategy 12).
2. Authority stays legible. A human may play directly, delegate a seat to an agent, or configure idle play. In every case the world records whose Will entered, through which tool and permission, rather than dissolving agency into anonymous automation. Power is not presumed benign or suspect: its source, asymmetry, exercise, resistance, and price must be readable.
3. Worlds remain durable and owned. The current product is local-first: the filesystem record is canon and IndexedDB is a rebuildable projection. Future shared Worlds must preserve explicit authority, portability, lineage, and custody rather than making a platform's hidden state the world.
4. Integrity compounds. One source of truth, no parallel paths, contracts over drift — in the product and in the codebase. Coherence beats cleverness because it's what lets the next change be cheap.
5. Growth by teaching. Bottom-up, through education — skills, diagrams, docs, the ladder. The concierge journey is a sales motion and is named as one; it still teaches, and the key things stay documented well.

How we work — the values turned inward (for every contributor, human or agent)

The same skepticism that is our product is our practice. If our whole reason to exist is that you can't trust confident output without verification, then we must not trust our own confident output either.

• Distrust the indicators. A green test, a passing build, a fluent model answer, a "this is the only caller" — these are indicators, not truth. The frontier-model failure mode is believing them. Verify the thing itself: read the code, run the check and read its output, prove the claim with a grep. This is the whole review-craft discipline, and it is a value, not just a technique.
• Enter humble. A new contributor reads before it writes. The code wins over any doc, any memory, any assumption. Inspect the current implementation; find the seam that already exists and extend it; don't fork a plausible parallel. Humility here isn't politeness — it's how you avoid being confidently wrong in a system you don't yet understand.
• Know when it's already good, and stop. Adding to something already under contract dilutes value and risks the shared tree. Finishing includes knowing you're finished.

New here? [.agents/skills/onboarding] walks you in — values first, then the docs your task needs.