Goals

Goals

Why this room exists

Palaestra needs scenarios more than it needs players.

A scenario is a day at the grid interface: a price series, a campus, a set of obligations, and a scoring function. The venue ships five. Five days is a curriculum, not a body of evidence. The interesting question the platform was built to answer is what happens when many people, in many regions, each compile the day they actually know something about, and then watch where the rules bend.

That requires a place where someone can take the machinery away, use it, and bring something back. Exedra is that place.

What Exedra is for

To let one person build a scenario end to end without asking permission. Clone, compile, play, verify, submit. No account, no cloud provider, no key, no conversation with a maintainer before the work starts.

To make the acceptance criteria inspectable in advance. The gate that judges a submission is in this repository and runs on your machine. You see the verdict before anyone else does, and you see it as many times as you like. A gate you cannot run is a gate you are guessing against.

To make the physics arguable. The engine is a few hundred lines of deterministic bookkeeping, and every rule it applies is written down. If the model is wrong, this is where you can show that it is wrong. The platform's stated posture is believable, not certified, and the checkable form of that claim is the list of what is not modeled.

To be forkable by an LLM. A contributor pointing a coding agent at this repository should find the context files, the prompts, and the worked examples that let the agent do useful work on the first pass rather than the fifth.

Who this is for

• A student or workshop team compiling their own region's day, publishing it under their own names, and defending it.
• An engineer with a grievance about the model. The most useful contribution may be a scenario that demonstrates the engine is missing something.
• Anyone who wants to read the machinery before deciding whether the venue's scores mean anything.

What success looks like

The library shelf at the venue is currently empty. The contribution mechanism runs end to end and has never been used by anyone outside the project. Success is the first accepted scenario that the maintainers did not write, followed by enough of them that the shelf is worth browsing.

A second measure matters more and is slower. Every scenario probes a standards-needs question: an obligation two reasonable operators would read differently, a rule that produces a perverse outcome, an ambiguity a player can exploit. A contributed scenario that surfaces one of those has done the thing the venue was built to do.

What Exedra is not

It is not the venue. It has no database, no leaderboard, no accounts, and no published standings. Those live at Palaestra and stay there. See What Exedra Withholds.

It is not a fork you are expected to maintain. The engine here is generated, and a pull request that edits it will be rejected on principle rather than on merit. Improvements to the engine belong upstream, where one copy governs every scored episode.

It is not a general-purpose grid simulator. It models energy bookkeeping at one seat, on purpose, and refuses to model the rest.