22 Commits

Author SHA1 Message Date
ea817d8044 fix(gui): Prevent page collapse by persistent config mount 2026-07-10 22:52:47 +05:30
382507e71a minor(gui): Disable simulation button if no provider found 2026-07-10 22:36:03 +05:30
7df685365e refactor(llm): Unify auto-bootstrapped prover instances and custom instances 2026-07-10 22:27:24 +05:30
b5fb48ed99 docs: refined readme 2026-07-10 17:08:55 +05:30
ae06982620 feat(gui): Switch to tailwind 2026-07-10 15:53:51 +05:30
d8d9015015 refactor: dynamically fetch available InferenceProviders for gui 2026-07-10 15:38:08 +05:30
Aditya Gupta
f88cc4efb4 docs: Improve Illustrations 2026-07-10 13:42:04 +05:30
185e68b541 feat: Added OpenRouter LLMProvider and setup bootstrapping 2026-07-10 07:59:31 +05:30
ebd0b76c23 minor: Updated favicon for web interfaces 2026-07-10 07:51:59 +05:30
2c842b1520 docs: Added CoC and Contributing guidelines 2026-07-10 07:51:48 +05:30
Aditya Gupta
b36517e5f3 Add files via upload 2026-07-10 07:44:14 +05:30
3c9c55157c merge: badge style change 2026-07-09 22:33:32 +05:30
cf6c015726 docs: Refined readme 2026-07-09 22:32:39 +05:30
8c0f1b45fd Update README.md 2026-07-09 22:16:20 +05:30
6d9155ef28 docs: Added docs link 2026-07-09 22:06:08 +05:30
b854dfe45c ci: fix node version 2026-07-09 21:57:45 +05:30
61c6fe8513 ci: Workflow for docs deployment to cf 2026-07-09 21:56:22 +05:30
30e26f78f9 minor: remove redundant pnpm scripts 2026-07-09 21:12:59 +05:30
1ebd5f77dc docs: Refine readme 2026-07-09 21:12:40 +05:30
Aditya Gupta
fd377e8794 Add files via upload 2026-07-09 20:53:30 +05:30
817bbde265 Merge pull request #21 from sortedcord/feat/config
Refactor LLM provider system and introduce GUI for simulations
2026-07-09 19:38:20 +05:30
4339a8b4b5 feat: Added openrouter provider 2026-07-09 11:13:02 +05:30
36 changed files with 2855 additions and 751 deletions

44
.github/workflows/deploy-docs.yml vendored Normal file
View File

@@ -0,0 +1,44 @@
name: Deploy Docs
on:
push:
branches:
- master
paths:
- 'web/docs/**'
- 'pnpm-lock.yaml'
- '.github/workflows/deploy-docs.yml'
workflow_dispatch:
jobs:
deploy:
runs-on: ubuntu-latest
name: Deploy Docs to Cloudflare Workers
steps:
- name: Checkout Repository
uses: actions/checkout@v4
- name: Setup pnpm
uses: pnpm/action-setup@v4
with:
version: 11
- name: Setup Node.js
uses: actions/setup-node@v4
with:
node-version: 22
cache: 'pnpm'
- name: Install Dependencies
run: pnpm install --frozen-lockfile
- name: Build Docs
run: pnpm --filter docs build
- name: Deploy to Cloudflare Workers
uses: cloudflare/wrangler-action@v3
with:
apiToken: ${{ secrets.CLOUDFLARE_API_TOKEN }}
accountId: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
workingDirectory: 'web/docs'

128
CODE_OF_CONDUCT.md Normal file
View File

@@ -0,0 +1,128 @@
# Contributor Covenant Code of Conduct
## Our Pledge
We as members, contributors, and leaders pledge to make participation in our
community a harassment-free experience for everyone, regardless of age, body
size, visible or invisible disability, ethnicity, sex characteristics, gender
identity and expression, level of experience, education, socio-economic status,
nationality, personal appearance, race, religion, or sexual identity
and orientation.
We pledge to act and interact in ways that contribute to an open, welcoming,
diverse, inclusive, and healthy community.
## Our Standards
Examples of behavior that contributes to a positive environment for our
community include:
- Demonstrating empathy and kindness toward other people
- Being respectful of differing opinions, viewpoints, and experiences
- Giving and gracefully accepting constructive feedback
- Accepting responsibility and apologizing to those affected by our mistakes,
and learning from the experience
- Focusing on what is best not just for us as individuals, but for the
overall community
Examples of unacceptable behavior include:
- The use of sexualized language or imagery, and sexual attention or
advances of any kind
- Trolling, insulting or derogatory comments, and personal or political attacks
- Public or private harassment
- Publishing others' private information, such as a physical or email
address, without their explicit permission
- Other conduct which could reasonably be considered inappropriate in a
professional setting
## Enforcement Responsibilities
Community leaders are responsible for clarifying and enforcing our standards of
acceptable behavior and will take appropriate and fair corrective action in
response to any behavior that they deem inappropriate, threatening, offensive,
or harmful.
Community leaders have the right and responsibility to remove, edit, or reject
comments, commits, code, wiki edits, issues, and other contributions that are
not aligned to this Code of Conduct, and will communicate reasons for moderation
decisions when appropriate.
## Scope
This Code of Conduct applies within all community spaces, and also applies when
an individual is officially representing the community in public spaces.
Examples of representing our community include using an official e-mail address,
posting via an official social media account, or acting as an appointed
representative at an online or offline event.
## Enforcement
Instances of abusive, harassing, or otherwise unacceptable behavior may be
reported to the community leaders responsible for enforcement at
mail@adityagupta.dev.
All complaints will be reviewed and investigated promptly and fairly.
All community leaders are obligated to respect the privacy and security of the
reporter of any incident.
## Enforcement Guidelines
Community leaders will follow these Community Impact Guidelines in determining
the consequences for any action they deem in violation of this Code of Conduct:
### 1. Correction
**Community Impact**: Use of inappropriate language or other behavior deemed
unprofessional or unwelcome in the community.
**Consequence**: A private, written warning from community leaders, providing
clarity around the nature of the violation and an explanation of why the
behavior was inappropriate. A public apology may be requested.
### 2. Warning
**Community Impact**: A violation through a single incident or series
of actions.
**Consequence**: A warning with consequences for continued behavior. No
interaction with the people involved, including unsolicited interaction with
those enforcing the Code of Conduct, for a specified period of time. This
includes avoiding interactions in community spaces as well as external channels
like social media. Violating these terms may lead to a temporary or
permanent ban.
### 3. Temporary Ban
**Community Impact**: A serious violation of community standards, including
sustained inappropriate behavior.
**Consequence**: A temporary ban from any sort of interaction or public
communication with the community for a specified period of time. No public or
private interaction with the people involved, including unsolicited interaction
with those enforcing the Code of Conduct, is allowed during this period.
Violating these terms may lead to a permanent ban.
### 4. Permanent Ban
**Community Impact**: Demonstrating a pattern of violation of community
standards, including sustained inappropriate behavior, harassment of an
individual, or aggression toward or disparagement of classes of individuals.
**Consequence**: A permanent ban from any sort of public interaction within
the community.
## Attribution
This Code of Conduct is adapted from the [Contributor Covenant][homepage],
version 2.0, available at
https://www.contributor-covenant.org/version/2/0/code_of_conduct.html.
Community Impact Guidelines were inspired by [Mozilla's code of conduct
enforcement ladder](https://github.com/mozilla/diversity).
[homepage]: https://www.contributor-covenant.org
For answers to common questions about this code of conduct, see the FAQ at
https://www.contributor-covenant.org/faq. Translations are available at
https://www.contributor-covenant.org/translations.

99
CONTRIBUTING.md Normal file
View File

@@ -0,0 +1,99 @@
# Contributing to Omnia
Thank you for your interest in contributing to Omnia! We welcome contributions from developers, technical writers, and anyone interested in agentic narrative simulation.
Please take a moment to review this document before submitting contributions.
## Table of Contents
1. [Documentation](#documentation)
2. [Getting Started](#getting-started)
3. [Development Workflow](#development-workflow)
4. [Coding Standards](#coding-standards)
5. [Pull Request Guidelines](#pull-request-guidelines)
## Documentation
The primary source of truth for the Omnia project is the official documentation:
👉 **[Omnia Documentation](https://omnia.adityagupta.dev/docs)**
Please refer to the documentation to understand the project architecture, memory model, spatial systems, intents framework, and custom LLM configurations.
## Getting Started
Omnia is organized as a monorepo managed with **pnpm** workspaces.
### Prerequisites
- **Node.js** (v22.13 or newer recommended)
- **pnpm** (v11 or newer recommended)
### Local Setup
1. Fork the repository and clone your fork:
```bash
git clone https://github.com/YOUR_USERNAME/omnia-consolidated.git
cd omnia-consolidated
```
2. Install dependencies:
```bash
pnpm install
```
3. Run the Web GUI interface locally:
```bash
pnpm dev:gui
```
4. Run the Starlight documentation site locally:
```bash
pnpm dev:docs
```
## Development Workflow
### Branching
Create a descriptive branch for your changes:
```bash
git checkout -b feature/your-feature-name
# or
git checkout -b fix/issue-description
```
### Running Tests
Make sure all unit tests pass before submitting changes:
```bash
# Run tests once
pnpm test
# Run tests in watch mode
pnpm test:watch
```
### Linting and Formatting
We enforce consistent code quality and formatting rules across the repository.
```bash
# Check code style and formatting
pnpm lint
pnpm format:check
# Auto-fix code style issues
pnpm lint:fix
pnpm format
```
## Coding Standards
- **TypeScript**: Omnia is written entirely in TypeScript. Ensure all new code is strongly typed.
- **Docstrings**: Document public-facing APIs, methods, and configurations.
## Pull Request Guidelines
1. **Keep PRs Focused**: Keep your changes as small and focused as possible.
2. **Include Tests**: If you are introducing a new feature or fixing a bug, write corresponding tests in `tests/`.
3. **Update Documentation**: If your changes alter public behavior or introduce new APIs, update the docs under `web/docs/src/content/docs/`.
4. **Follow Commit Conventions**: Write clear, descriptive commit messages.

139
README.md
View File

@@ -1,49 +1,97 @@
![Omnia Logo](web/docs/src/assets/img/logo.png)
<p align="center">
<img src="web/docs/src/assets/img/logo.png" alt="Omnia Logo" />
</p>
An LLM-assisted narrative simulation engine where the <b>world state lives outside the model</b>, characters act through <b>intents that get validated</b> and applied by engine code, and each character's knowledge, memory, and emotional state are subjective and partial by construction.
<h1 align="center">Omnia</h1>
Omnia is an engine for building narrative RPG-style worlds where characters are played by a language model. It is built to survive long play sessions instead of falling apart after twenty minutes.
<p align="center">
<b>An architectural framework for multi agent-narrative simulations and fictional worlds!</b>
</p>
<p align="center">
<a href="https://omnia.adityagupta.dev/docs"><img src="https://img.shields.io/badge/Omnia_Docs-Read_The_Docs-red?style=for-the-badge" alt="Docs" /></a>
<img src="https://img.shields.io/github/license/sortedcord/omnia-consolidated?style=for-the-badge" alt="License" />
<img src="https://img.shields.io/github/repo-size/sortedcord/omnia-consolidated?style=for-the-badge" alt="Repo Size" />
<img src="https://img.shields.io/github/languages/top/sortedcord/omnia-consolidated?style=for-the-badge" alt="Top Language" />
</p>
## The Problem with the Naive Approach
The <b>world state lives outside the model</b>, characters act through <b>intents that get validated</b> and applied by engine code. Each character's knowledge, memory, and emotional state are subjective and partial by construction.
Single-agent, single-context systems (AI Dungeon and its descendants) prompt one model to _be_ the world and everyone in it. That breaks in predictable ways over long sessions:
<p align="center">
<img src="./web/docs/src/assets/img/puppet.webp" />
</p>
- **State Leaks:** Characters know things they had no way of learning, because a model with full context cannot help but use it. The assassin's target greets him by name.
- **Secrets Refuse to Stay Secret:** "Don't reveal this" is a suggestion a model can argue past, not a mechanism that says no. One clever player question and the conspiracy folds.
Single-agent or single-context systems (AI Dungeon and its descendants) prompt one model to _be_ the world and everyone in it. That breaks in predictable ways over long sessions:
- **State Leaks:** Characters know things they had no way of learning, because a model with full context cannot help but use it.
- **Consequences Evaporate:** Betray someone, apologize, and they forgive you a turn later because nothing is tracking the betrayal as a persistent fact.
- **Emotional Drift:** Emotional state is either frozen into a meaningless number (`trust: 40`) or handed to the model to grade itself, producing drifting, arbitrary values.
- **Stat Drift:** Statistical attributes are either frozen into a meaningless number (`trust: 40`) or handed to the model to grade itself, producing drifting, arbitrary values.
- **World Rot:** The world state slowly contradicts itself because the model has no structured place to keep it. The locked door is open, then locked, then never existed.
- **Everyone Is One Person:** Every character shares one context, so every character shares one mind. They can't genuinely surprise each other, lie to each other, or know different things — they're sock puppets on the same hand.
- **Everyone Is One Person:** Every character shares one context, so every character shares one mind. They can't genuinely surprise each other, lie to each other, or know different things. They're sock puppets on the hands of one puppetmaster.
The root cause is the same in every case: the model is being asked to be the database, the physics engine, the referee, and the whole cast simultaneously inside a context window that forgets, blends, and leaks.
The model should not be the database, the physics engine and the whole cast simultaneously inside a sliding context window.
## The Omnia Solution
Omnia answers every one of these failures with the same move: pull the thing that has to stay consistent out of the model and into structured, queryable, code-controlled state.
Omnia answers every one of these failures with the same move: **pull the thing that has to stay consistent out of the model** and into structured, queryable, code-controlled state.
- **World State:** Lives in a SQLite database, not in a context window. It cannot drift, because nothing regenerates it — it only changes through validated deltas.
- **World State:** Lives in a DB, not a context window. It cannot drift, because nothing regenerates it. The world state only changes through validated deltas.
- **Actions:** Actions are proposals (Intents) that engine code validates and applies; they are never direct edits the model makes to the world. The model proposes; deterministic code disposes.
- **Epistemic Privacy:** Knowledge, memory, and emotion are modeled per character and kept partial on purpose. A character literally cannot reach for what it has not earned the right to know — the secret is not in its prompt, so there is nothing to jailbreak out of it.
- **Epistemic Privacy:** Knowledge, memory, and emotion are modeled per character and kept partial on purpose. A character literally cannot reach for what it has not earned the right to know. The secret is not in its prompt, so there is **nothing to jailbreak out of it**.
## What This Buys You
## What this buys you
The payoff is scenario complexity that uni-agent systems structurally cannot represent, no matter how good the model gets:
<p align="center">
<img src="./web/docs/src/assets/img/features.webp" />
</p>
The payoff is scenario complexity that **uni-agent systems structurally cannot represent, no matter how good the model gets**.
- **Real secrets, real dramatic irony.** One NPC knows the sword is cursed; the other does not. This holds for hundreds of turns not because the model is disciplined, but because the second NPC's prompts are constructed from an attribute set that simply does not contain the fact. Leaking it would require the engine to have handed it over.
- **Genuine deception between characters.** Because each character acts from its own bounded view, characters can lie to each other and be believed with the truth intact in the world state. A con game, a mole in the party, an unreliable ally: these are queries over who-knows-what, not prompt acrobatics.
- **Betrayal that stays betrayed.** Events persist as per-observer memory entries with outcomes. An apology adds a memory; it does not delete one.
- **Divergent accounts of the same event.** Two witnesses to the same scene hold two different buffer entries, filtered through their own aliases and vantage points. Ask them separately what happened and you get testimony, not a transcript.
- **Identity as information.** Characters refer to each other through subjective alias maps ("the hooded figure" vs. "Bob"). Recognizing someone, being recognized, or staying anonymous are all mechanical states — a masked stranger is a masked stranger until the engine says otherwise.
- **A physics referee that can say no.** "I pick the lock with a hairpin" is validated against world state by the Architect before anything changes. Failure is a recorded outcome the character remembers, not a narrative the model politely retconned.
- **Time that behaves.** A world clock advances by validated, per-action deltas, and memory is recalled with psychologically natural phrasing ("earlier today, in the afternoon" — not a timestamp). Long timelines stay coherent because time is data, not vibes.
- **Genuine deception between characters.** Because each entity acts from its own bounded view, they can lie to each other and be believed; with the truth intact in the world state. A con game, a mole in the party, an unreliable ally: these are queries over "who knows what" and not prompt engineering.
- Events persist as per observer memory entries with outcomes. An apology adds a memory; it does not delete one.
- **Divergent accounts of the same event.** Two witnesses to the same scene hold two different buffer entries, filtered through their own aliases and vantage points. Ask them separately what happened and you get varied testimony.
- **A physics referee that can say no.** `I pick the lock with a hairpin` is validated against world state by the Architect before anything changes. Failure is a recorded outcome the entity remembers.
- **Time that behaves.** A world clock advances by validated, per-action deltas, and memory is recalled with psychologically natural phrasing ("earlier today, in the afternoon" — not a timestamp). `TimeOfDay` is deterministic and not based on vibes.
- **No main character syndrome.** The simulation runs fully autonomously or you act on behalf of any entity. You, the player, are just an entity in the data model, not structurally elevated above the rest of the world. The world can exist without the you.
- **Granular model control.** Omnia is not locked to a single LLM. You can pick a different model for **every individual step that calls the LLM**. Narration prose that demands richer reasoning gets a frontier model; quick intent decoding or other generators get smaller ones or a model running entirely on your local machine.
The general principle: **anything that must remain true is state; the model only ever supplies behavior.** That division of labor is what lets the cast, the secrets, and the timeline scale without the fiction collapsing.
The general principle: **anything that must remain true is state; the model only ever supplies behavior.**
## Installation
### Prerequisites
- [Node.js](https://nodejs.org/) (v20+ recommended)
- [pnpm](https://pnpm.io/) (v9+ recommended)
- An API key for Google Gemini (`GOOGLE_API_KEY` environment variable), or configured settings via the GUI.
### Installation
1. Clone the repository:
```bash
git clone https://github.com/sortedcord/omnia-consolidated.git
cd omnia-consolidated
```
2. Install dependencies:
```bash
pnpm install
```
### Running the Web GUI
To launch the Next.js development server for the GUI dashboard:
```bash
pnpm dev:gui
```
Access the application locally at `http://localhost:3000`.
## Core Architecture
### The Actor Agent
Each character takes turns through an **Actor Agent** that receives a strictly epistemically-bounded prompt: its own attributes (public, plus private ones explicitly granted to itself), its subjective memory buffer, the entities co-present at its location, and the current moment. Nothing else. The actor responds with free narrative prose — what the character does, says, or _thinks_.
Each entity takes turns through an **Actor Agent** that receives a strictly epistemically bounded prompt: its own attributes (public, plus private ones explicitly granted to itself), its subjective memory buffer, the entities co-present at its location, and the current moment. Nothing else. The actor responds with free narrative prose.
Prose is decoded into typed intents:
@@ -51,29 +99,29 @@ Prose is decoded into typed intents:
- **`action`** — a physical act, subject to validation.
- **`monologue`** — an inner thought. No one else perceives it, it bypasses validation entirely, and it is written straight into the character's private memory.
Not every turn needs an outward act; a character may simply think. This is what makes characters feel inhabited rather than reactive and it produces a durable, queryable record of each character's private reasoning (see [Research Instrument](#a-research-instrument-model-psychology-in-fiction) below).
Not every turn needs an outward act; a character may simply think. This is what makes characters feel inhabited rather than reactive and it produces a queryable record of each character's private reasoning (see [Research Instrument](#a-research-instrument-model-psychology-in-fiction) below).
The prose generator is pluggable (`IActorProseGenerator`): the same turn loop runs an LLM-driven NPC or a human at a CLI prompt, identically bounded by what their character knows.
The prose generator is pluggable (`IActorProseGenerator`): the same turn loop runs an LLM driven NPC or a human, identically bounded by what their character knows. (This is what eliminates Main Character Syndrome)
### Intents & The World Architect
An action becomes an **Intent** — a cheap, declarative, allowed-to-be-wrong proposal. Intents route to the **World Architect**, which validates them against the objective world state (dialogue is exempt; monologue never even arrives) and generates structured deltas — starting with time advancement — that deterministic code applies after strict schema (Zod) validation. The model proposes a change; it never touches the database.
An action becomes an **Intent** which is a simple proposal that is _allowed to be wrong_. Intents route to the **World Architect**, which validates them against the objective world state and generates structured deltas like time advancement, attribute change, etc. This deterministic code applies after strict schema (Zod) validation. The model never touches the DB.
This is the load-bearing wall. Because every mutation flows through one validated chokepoint, the world cannot rot: there is no second copy of reality inside a context window to fall out of sync.
Because every mutation flows through one validated chokepoint, the world cannot rot: there is no second copy of reality inside a context window to fall out of sync.
### Attribute-Level Privacy
### Attribute Level Privacy
Every entity, item, and location is an attribute bag. Each attribute carries its own visibility (`PUBLIC` or `PRIVATE`) with an explicit access list. "The sword is cursed" is a private attribute checked in code, not a rule the model is politely asked to honor. Privacy lives at the level of the fact, not the entity — a character can be publicly a blacksmith and privately a spy, and even facts about _itself_ are hidden from it unless explicitly granted (amnesia, repression, and unwitting sleeper agents come free with the model).
Every entity, item, and location is an _attribute bag_. Each attribute carries its own visibility (`PUBLIC` or `PRIVATE`) with an explicit access list. "The sword is cursed" is a private attribute checked in code, not a rule the model is politely asked to honor. Privacy lives at the level of the fact, not the entity. A character can be publicly a blacksmith and privately a spy, and even facts about _itself_ are hidden from it unless explicitly granted (amnesia, repression, and unwitting sleeper agents come free with the model).
The dividend: **prompt-injection-proof secrets.** There is no instruction to override because the information was never serialized into the prompt. Epistemic privacy turns "the model shouldn't say this" (hard, unreliable) into "the model doesn't know this" (trivial, absolute).
### Spatial Perception
Space is a graph: `world → region → location → point of interest`, connected by portals with sound and vision propagation values. When something happens, it bubbles outward. There are no coordinates, no pathfinding, no collision geometry — a narrative engine doesn't need a tactical simulation, and a discrete graph is sufficient. Today actors perceive co-located entities and their location's visible attributes; portal-propagated perception is on the roadmap.
Space is a graph: `world → region → location → point of interest`, connected by portals with sound and vision propagation values. When something happens, it bubbles outward. There are no coordinates, no pathfinding, no collision geometry — a narrative engine doesn't need a tactical simulation, and a discrete graph is sufficient. Today actors perceive co-located entities and their location's visible attributes; portal propagated perception is on the roadmap.
### Memory Tiers
- **Verbatim Buffer (implemented):** Per-character subjective event log. Every entry is stored from the owner's perspective actors resolved through the owner's alias map, outcomes attached — and recalled with naturalized time phrasing.
- **Verbatim Buffer (implemented):** Per-character subjective event log. Every entry is stored from the owner's perspective actors resolved through the owner's alias map, outcomes attached — and recalled with naturalized time phrasing.
- **Vector Archive (planned):** Summarized, embedded memory entries for semantic retrieval, keeping verbatim quotes only for high-salience lines.
- **Dossier (planned):** Each observer's subjective beliefs about another character.
@@ -83,16 +131,20 @@ Memory is per-character on purpose: recall is testimony from a vantage point, wh
Rather than a scalar the model drifts, every significant interaction becomes a ledger entry with an affect vector across OCC-derived dimensions (plus arousal, dominance, and social drive). The model judges a single moment; deterministic code aggregates the ledger over time with decay and attention weighting. A character can be simultaneously furious about one thing and grateful for another, and an apology does not silently erase a betrayal.
This however is something that I haven't implementing or plan to implement anytime soon. The mathematical models described in NLAVS is still very abstract and subject to a lot of changes. CAA and RepE is still cutting edge research that I'm still reading papers about.
Omnia might get an affect vector system however, it's going to be more simplistic than what the NLAVS proposal scribbles down.
## A Research Instrument: Model Psychology in Fiction
Omnia's architecture doubles as an apparatus for studying how language models behave _as characters_ under controlled epistemic conditions — something uni-agent setups cannot do, because they can neither control what the model knows nor observe what it withholds.
Omnia's architecture doubles as an apparatus for studying how language models behave _as characters_ under controlled epistemic conditions.
- **A window into private reasoning.** Monologue intents are the model's in-character thoughts: unperceived by other agents, exempt from validation, but durably logged. You can directly compare what a character _thinks_ against what it _says and does_ measuring deception, self-consistency, motivated reasoning, or the gap between private appraisal and public behavior.
- **Knowledge as an experimental variable.** Attribute ACLs let you administer information with precision: give one agent a fact, withhold it from another, and observe propagation, inference, and leakage through dialogue alone. Secret-keeping stops being anecdotal and becomes testable — _provably_, since the engine logs exactly what each agent was ever shown.
- **Controlled, reproducible conditions.** A scenario is a JSON file; a run is a SQLite database. Identical initial conditions, swappable model providers behind one interface (`ILLMProvider`), and a deterministic mock for baselines. Rerun the white-room experiment a hundred times, vary one attribute, and diff the transcripts.
- **Multi-agent social dynamics with ground truth.** Because objective world state exists independently of any agent's beliefs, you can score agents' beliefs and claims against reality hallucination, confabulation, and social conformity become measurable quantities rather than impressions.
- **A window into private reasoning.** Monologue intents are the model's in character thoughts: unperceived by other agents, exempt from validation, but durably logged. You can directly compare what a character _thinks_ against what it _says and does_: measuring deception, self-consistency, motivated reasoning, etc.
- **Knowledge as an experimental variable.** Attribute ACLs let you administer information with precision: give one agent a fact, withhold it from another, and observe propagation, inference, and leakage through dialogue alone.
- **Controlled, reproducible conditions.** A scenario is a JSON file (like a template); a run is a SQLite database. Identical initial conditions, swappable model providers behind one interface (`ILLMProvider`), and a deterministic mock for baselines. Rerun the scenario a hundred times, vary one attribute, and diff the transcripts.
- **Multi agent social dynamics with ground truth.** Because objective world state exists independently of any agent's beliefs, you can score agents' beliefs and claims against reality like hallucination. Even social conformity become measurable quantities rather than impressions or _✨ vibes_.
The bundled demo scenario is exactly this: [`talking-room`](./content/demo/scenarios/talking-room.json) places two memory-wiped subjects in a featureless white room — each knowing their own name but not the other's — and observes what they do. It runs today, via the CLI, with a human optionally playing either subject.
The bundled demo scenario is exactly this: [`talking-room`](./content/demo/scenarios/talking-room.json) places two memory wiped subjects in a featureless white room. Each know their own name but not the other's. Observe what they do. It runs today, ~~via the CLI, with a human optionally playing either subject~~ via a GUI which is in rapid development. You can let it run forever autonomously or roleplay as either character.
## Project Status: What `v0` Means
@@ -112,15 +164,15 @@ The finish line for the first milestone is small on purpose.
- [x] Two hand-authored NPCs live in one location, playable via CLI.
- [ ] Each has buffer and vector-archive memory and recalls something said a few turns earlier. _(buffer: done; vector archive: not started)_
- [ ] One NPC knows a fact the other does not and, provably by testing, will not leak it.
- [x] One NPC knows a fact the other does not and, provably by testing, will not leak it.
- [x] The Architect processes at least one non-trivial action per exchange with a visible state change.
- [x] The whole thing persists to a SQLite file and reloads identically.
**Explicitly out of scope for `v0`:** Constraint validators (beyond basic sense-checking), multi-location perception, affect-vector decay math, the Dossier, whims/simulation tiering, the delta ledger, and UI beyond CLI.
**Explicitly out of scope for `v0`:** Constraint validators (beyond basic sense-checking), multi-location perception, affect-vector decay math, the Dossier, whims/simulation tiering, the delta ledger.
### A Note on Tech Debt
The Architect currently trusts an LLM's judgement about reasonable consequences rather than validating every change against declarative constraints. A general constraint solver is worth building eventually, but building it before anything is playable is foundational perfectionism that produces beautiful architecture and no game. `v0` keeps the single-call Architect on purpose.
The Architect currently trusts an LLM's judgement about reasonable consequences rather than validating every change against declarative constraints. A general constraint solver is worth building eventually, but building it before anything is playable is foundational perfectionism that produces beautiful architecture and no framework. `v0` keeps the single-call Architect on purpose.
## Repository Layout
@@ -138,17 +190,16 @@ omnia/
llm/ ILLMProvider interface plus Gemini and deterministic mock implementations
scenario/ scenario JSON schema and loader (JSON → SQLite)
apps/
cli/ the playable loop (human or LLM actors, --scenario / --play flags)
gui/ Next.js Web GUI dashboard and simulation runner
content/
demo/ bundled scenarios (talking-room)
tests/
integration/ cross-package tests against a mocked LLM
evals/ deliberate real-API evaluation runs
docs/ Astro documentation site (→ web/docs/)
web/
docs/ Astro documentation site
```
_The engine core deliberately knows nothing about domain content (stats, traits, genres). Scenarios are plain JSON the loader ingests; what an attribute means is the scenario's business, not the engine's._
## Roadmap (Build Order after `v0`)
1. Vector-archive memory and retrieval (closing out the `v0` memory milestone).

View File

@@ -1,6 +1,6 @@
/// <reference types="next" />
/// <reference types="next/image-types/global" />
import "./.next/types/routes.d.ts";
import "./.next/dev/types/routes.d.ts";
// NOTE: This file should not be edited
// see https://nextjs.org/docs/app/api-reference/config/typescript for more information.

View File

@@ -10,14 +10,14 @@
"lint": "next lint"
},
"dependencies": {
"@omnia/core": "workspace:*",
"@omnia/llm": "workspace:*",
"@omnia/intent": "workspace:*",
"@omnia/architect": "workspace:*",
"@omnia/actor": "workspace:*",
"@omnia/architect": "workspace:*",
"@omnia/core": "workspace:*",
"@omnia/intent": "workspace:*",
"@omnia/llm": "workspace:*",
"@omnia/memory": "workspace:*",
"@omnia/spatial": "workspace:*",
"@omnia/scenario": "workspace:*",
"@omnia/spatial": "workspace:*",
"dotenv": "^17.4.2",
"next": "^16.2.10",
"react": "^19.2.0",
@@ -27,6 +27,9 @@
"@types/node": "^26.1.0",
"@types/react": "^19.2.0",
"@types/react-dom": "^19.2.0",
"autoprefixer": "^10.5.2",
"postcss": "^8.5.16",
"tailwindcss": "^3.4.19",
"typescript": "^6.0.3"
}
}

View File

@@ -0,0 +1,8 @@
const config = {
plugins: {
tailwindcss: {},
autoprefixer: {},
},
};
export default config;

View File

@@ -10,8 +10,9 @@ import {
getProviderMappings,
setProviderMapping,
updateProviderInstance,
getAvailableProviders,
} from "@/app/play/actions";
import type { LLMProviderInstance } from "@omnia/llm";
import type { LLMProviderInstance, LLMProviderMeta } from "@omnia/llm";
interface ConfigStatus {
apiKeySet: boolean;
@@ -24,10 +25,15 @@ export default function ConfigPage() {
const [config, setConfig] = useState<ConfigStatus | null>(null);
const [instances, setInstances] = useState<LLMProviderInstance[]>([]);
const [mappings, setMappings] = useState<Record<string, string>>({});
const [availableProviders, setAvailableProviders] = useState<
LLMProviderMeta[]
>([]);
const [loading, setLoading] = useState(true);
const [error, setError] = useState("");
const [selectedInstanceId, setSelectedInstanceId] = useState<string | "new">("new");
const [selectedInstanceId, setSelectedInstanceId] = useState<string | null>(
null,
);
const [editName, setEditName] = useState("");
const [editProvider, setEditProvider] = useState("google-genai");
const [editKey, setEditKey] = useState("");
@@ -35,23 +41,44 @@ export default function ConfigPage() {
const [editIsActive, setEditIsActive] = useState(false);
useEffect(() => {
if (selectedInstanceId === "new") {
if (selectedInstanceId === null) {
setEditName("");
setEditProvider("google-genai");
setEditKey("");
setEditModel("gemini-2.5-flash");
setEditIsActive(false);
} else if (selectedInstanceId === "new") {
setEditName("");
const defaultProvider = "google-genai";
setEditProvider(defaultProvider);
setEditKey("");
const pMeta = availableProviders.find((p) => p.id === defaultProvider);
setEditModel(pMeta?.defaultModel || "gemini-2.5-flash");
setEditIsActive(false);
} else {
const inst = instances.find((i) => i.id === selectedInstanceId);
if (inst) {
setEditName(inst.name);
setEditProvider(inst.providerName);
setEditKey("");
setEditModel(inst.modelName || "gemini-2.5-flash");
const pMeta = availableProviders.find(
(p) => p.id === inst.providerName,
);
setEditModel(
inst.modelName || pMeta?.defaultModel || "gemini-2.5-flash",
);
setEditIsActive(inst.isActive);
}
}
}, [selectedInstanceId, instances]);
}, [selectedInstanceId, instances, availableProviders]);
const handleProviderChange = (providerId: string) => {
setEditProvider(providerId);
const pMeta = availableProviders.find((p) => p.id === providerId);
if (pMeta) {
setEditModel(pMeta.defaultModel);
}
};
const loadInstances = useCallback(async () => {
try {
@@ -79,6 +106,8 @@ export default function ConfigPage() {
setConfig(result);
await loadInstances();
await loadMappings();
const provs = await getAvailableProviders();
setAvailableProviders(provs);
} catch (err) {
setError(err instanceof Error ? err.message : String(err));
} finally {
@@ -107,13 +136,24 @@ export default function ConfigPage() {
setLoading(false);
return;
}
const created = await createProviderInstance(editName, editProvider, editKey, editModel || undefined);
const created = await createProviderInstance(
editName,
editProvider,
editKey,
editModel || undefined,
);
if (editIsActive) {
await setActiveProviderInstance(created.id);
}
setSelectedInstanceId(created.id);
} else {
await updateProviderInstance(selectedInstanceId, editName, editProvider, editKey || undefined, editModel || undefined);
await updateProviderInstance(
selectedInstanceId,
editName,
editProvider,
editKey || undefined,
editModel || undefined,
);
if (editIsActive) {
await setActiveProviderInstance(selectedInstanceId);
}
@@ -129,14 +169,15 @@ export default function ConfigPage() {
};
const handleDelete = async () => {
if (selectedInstanceId === "new") return;
if (!confirm("Are you sure you want to delete this provider instance?")) return;
if (selectedInstanceId === "new" || selectedInstanceId === null) return;
if (!confirm("Are you sure you want to delete this provider instance?"))
return;
try {
setLoading(true);
setError("");
await deleteProviderInstance(selectedInstanceId);
setSelectedInstanceId("new");
setSelectedInstanceId(null);
await loadInstances();
await loadMappings();
} catch (err) {
@@ -146,7 +187,10 @@ export default function ConfigPage() {
}
};
const handleUpdateMapping = async (task: string, providerInstanceId: string) => {
const handleUpdateMapping = async (
task: string,
providerInstanceId: string,
) => {
try {
setLoading(true);
await setProviderMapping(task, providerInstanceId);
@@ -159,44 +203,61 @@ export default function ConfigPage() {
};
return (
<div className="config-page">
<h1>Configuration</h1>
<div className="mx-auto max-w-[800px] px-4 py-8">
<h1 className="mb-6 text-2xl">Configuration</h1>
{loading && <p>Loading configuration...</p>}
{error && <div className="error-banner">{error}</div>}
{config === null && loading && <p>Loading configuration...</p>}
{error && (
<div className="mb-4 rounded border border-red-300 bg-red-50 px-3 py-2 text-sm text-red-700">
{error}
</div>
)}
{config && !loading && (
<>
<section className="config-section">
<h2>LLM Provider Instances</h2>
<div className="provider-split-container">
{config && (
<div className={loading ? "opacity-60 pointer-events-none transition-opacity duration-200" : "transition-opacity duration-200"}>
<section className="mb-8 pb-6">
<h2 className="mb-3 text-lg">LLM Provider Instances</h2>
<div className="mt-4 grid min-h-[400px] grid-cols-1 overflow-hidden rounded-xl border border-gray-200 bg-white md:grid-cols-[30%_70%]">
{/* 30% area */}
<div className="provider-list-pane">
<div className="pane-header">
<h3>Instances</h3>
<div className="flex flex-col border-r border-gray-200 bg-gray-50">
<div className="flex items-center justify-between border-b border-gray-200 bg-gray-100 px-4 py-4">
<h3 className="m-0 text-[0.95rem] font-semibold text-[#111]">
Instances
</h3>
<button
onClick={() => setSelectedInstanceId("new")}
className="btn-add-inst"
className="cursor-pointer rounded-md bg-emerald-500 px-3 py-1.5 text-xs font-medium text-white transition-colors hover:bg-emerald-600"
type="button"
>
+ Add
</button>
</div>
<div className="pane-list">
<div className="flex flex-1 flex-col overflow-y-auto">
{instances.length === 0 ? (
<div className="no-instances-msg">No instances configured</div>
<div className="px-4 py-8 text-center text-xs text-gray-400">
No instances configured
</div>
) : (
instances.map((inst) => (
<div
key={inst.id}
onClick={() => setSelectedInstanceId(inst.id)}
className={`instance-list-item ${selectedInstanceId === inst.id ? "active" : ""}`}
className={`cursor-pointer border-b border-gray-200 border-l-[3px] px-4 py-4 transition-all hover:bg-gray-100 ${
selectedInstanceId === inst.id
? "border-l-blue-500 bg-blue-50"
: "border-l-transparent"
}`}
>
<div className="item-name">{inst.name}</div>
<div className="item-meta">
<div className="text-sm font-medium text-[#111]">
{inst.name}
</div>
<div className="mt-1 flex items-center justify-between text-xs text-gray-500">
<span>{inst.providerName}</span>
{inst.isActive && <span className="active-pill">Active</span>}
{inst.isActive && (
<span className="rounded-full bg-green-100 px-1.5 py-[1px] text-[0.65rem] font-semibold text-green-700">
Active
</span>
)}
</div>
</div>
))
@@ -205,127 +266,208 @@ export default function ConfigPage() {
</div>
{/* 70% area */}
<div className="provider-form-pane">
<form onSubmit={handleSave} className="provider-config-form">
<div className="form-scroll-content">
<h3>
{selectedInstanceId === "new"
? "Create New Provider Instance"
: `Configure: ${editName}`}
</h3>
<div className="form-group">
<label htmlFor="formName">Friendly Name</label>
<input
id="formName"
type="text"
value={editName}
onChange={(e) => setEditName(e.target.value)}
placeholder="e.g. Gemini - Production"
required
/>
</div>
<div className="form-group">
<label htmlFor="formProvider">Provider Type</label>
<select
id="formProvider"
value={editProvider}
onChange={(e) => setEditProvider(e.target.value)}
>
<option value="google-genai">Google Gemini (Gemini-2.5-flash)</option>
<option value="mock">Mock LLM Provider</option>
</select>
</div>
<div className="form-group">
<label htmlFor="formKey">API Key</label>
<input
id="formKey"
type="password"
value={editKey}
onChange={(e) => setEditKey(e.target.value)}
placeholder={
selectedInstanceId === "new"
? "AIzaSy..."
: "•••••••• (unchanged)"
}
required={selectedInstanceId === "new"}
/>
</div>
<div className="form-group">
<label htmlFor="formModel">Model Name</label>
<input
id="formModel"
type="text"
value={editModel}
onChange={(e) => setEditModel(e.target.value)}
placeholder="e.g. gemini-2.5-flash, gemini-2.5-pro"
/>
</div>
<div className="form-group checkbox-group">
<input
id="formActive"
type="checkbox"
checked={editIsActive}
onChange={(e) => setEditIsActive(e.target.checked)}
/>
<label htmlFor="formActive">Set as Active Instance</label>
</div>
<div className="flex flex-col bg-white">
{selectedInstanceId === null ? (
<div className="flex flex-1 flex-col items-center justify-center p-6 text-center text-sm text-gray-400">
Press + to add or select an existing Instance to edit
</div>
) : (
<form
onSubmit={handleSave}
className="flex h-full flex-col justify-between"
>
<div className="flex flex-1 flex-col gap-5 p-6">
<h3 className="m-0 mb-2 text-lg font-semibold text-[#111]">
{selectedInstanceId === "new"
? "Create New Provider Instance"
: `Configure: ${editName}`}
</h3>
<div className="form-actions-bar">
<div className="action-left">
{selectedInstanceId !== "new" && (
<button
type="button"
onClick={handleDelete}
disabled={loading}
className="btn-delete-pane"
<div className="flex flex-col gap-1.5">
<label
htmlFor="formName"
className="text-xs font-medium text-gray-700"
>
Delete
</button>
)}
</div>
<div className="action-right">
<button type="submit" disabled={loading} className="btn-save-pane">
{loading ? "Saving..." : "Save"}
</button>
</div>
</div>
</form>
</div>
Friendly Name
</label>
<input
id="formName"
type="text"
value={editName}
onChange={(e) => setEditName(e.target.value)}
placeholder="e.g. Gemini - Production"
required
className="w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm outline-none transition-[border-color,box-shadow] focus:border-blue-500 focus:ring-3 focus:ring-blue-500/15"
/>
</div>
<div className="flex flex-col gap-1.5">
<label
htmlFor="formProvider"
className="text-xs font-medium text-gray-700"
>
Provider Type
</label>
<select
id="formProvider"
value={editProvider}
onChange={(e) => handleProviderChange(e.target.value)}
className="w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm outline-none transition-[border-color,box-shadow] focus:border-blue-500 focus:ring-3 focus:ring-blue-500/15"
>
{availableProviders.map((p) => (
<option key={p.id} value={p.id}>
{p.displayName}
</option>
))}
</select>
{editProvider && availableProviders.length > 0 && (
<span className="mt-1 block rounded border border-gray-200 bg-gray-100 px-3 py-2 text-xs text-gray-600">
{
availableProviders.find(
(p) => p.id === editProvider,
)?.description
}
</span>
)}
</div>
<div className="flex flex-col gap-1.5">
<label
htmlFor="formKey"
className="text-xs font-medium text-gray-700"
>
API Key
</label>
<input
id="formKey"
type="password"
value={editKey}
onChange={(e) => setEditKey(e.target.value)}
placeholder={
selectedInstanceId === "new"
? "AIzaSy..."
: "•••••••• (unchanged)"
}
required={selectedInstanceId === "new"}
className="w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm outline-none transition-[border-color,box-shadow] focus:border-blue-500 focus:ring-3 focus:ring-blue-500/15"
/>
</div>
<div className="flex flex-col gap-1.5">
<label
htmlFor="formModel"
className="text-xs font-medium text-gray-700"
>
Model Name
</label>
<input
id="formModel"
type="text"
value={editModel}
onChange={(e) => setEditModel(e.target.value)}
placeholder="e.g. gemini-2.5-flash, gemini-2.5-pro"
className="w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm outline-none transition-[border-color,box-shadow] focus:border-blue-500 focus:ring-3 focus:ring-blue-500/15"
/>
</div>
<div className="mt-1 flex flex-row items-center gap-2">
<input
id="formActive"
type="checkbox"
checked={editIsActive}
onChange={(e) => setEditIsActive(e.target.checked)}
className="h-4 w-4 cursor-pointer"
/>
<label
htmlFor="formActive"
className="cursor-pointer text-xs font-medium text-gray-700"
>
Set as Active Instance
</label>
</div>
</div>
<div className="flex items-center justify-between border-t border-gray-200 bg-gray-50 px-6 py-4">
<div>
{selectedInstanceId !== "new" && (
<button
type="button"
onClick={handleDelete}
disabled={loading}
className="cursor-pointer rounded-md bg-red-500 px-4 py-2 text-sm font-medium text-white transition-colors hover:bg-red-600 disabled:opacity-50"
>
Delete
</button>
)}
</div>
<div>
<button
type="submit"
disabled={loading}
className="cursor-pointer rounded-md bg-blue-600 px-5 py-2 text-sm font-medium text-white transition-colors hover:bg-blue-700 disabled:opacity-50"
>
{loading ? "Saving..." : "Save"}
</button>
</div>
</div>
</form>
)}
</div>
</div>
</section>
<section className="config-section">
<h2>Task Provider Routing</h2>
<p className="config-hint" style={{ background: "#eff6ff", border: "1px solid #bfdbfe", color: "#1e3a8a", margin: "1rem 0" }}>
Configure which LLM Provider Key Instance should handle each specific simulation task. Mappings default to the currently <strong>Active</strong> instance if not specified.
<section className="mb-8 pb-6">
<h2 className="mb-3 text-lg">Task Provider Routing</h2>
<p className="my-4 rounded border border-blue-200 bg-blue-50 px-3 py-2 text-xs text-blue-800">
Configure which LLM Provider Key Instance should handle each
specific simulation task. Mappings default to the currently{" "}
<strong>Active</strong> instance if not specified.
</p>
<div className="mappings-grid">
<div className="mt-4 grid grid-cols-1 gap-4 md:grid-cols-2">
{[
{ key: "actor-prose", label: "Actor Prose Generation", desc: "Generates roleplay/narrative prose for Non-Player Characters." },
{ key: "llm-validator", label: "LLM Validator", desc: "Arbitrates and validates proposed actions against the world state rules." },
{ key: "intent-decoder", label: "Intent Decoder", desc: "Splits raw prose actions into structured intents (Player and NPC)." },
{ key: "timedelta", label: "TimeDelta Generator", desc: "Calculates the duration of character actions to advance the game clock." },
{
key: "actor-prose",
label: "Actor Prose Generation",
desc: "Generates roleplay/narrative prose for Non-Player Characters.",
},
{
key: "llm-validator",
label: "LLM Validator",
desc: "Arbitrates and validates proposed actions against the world state rules.",
},
{
key: "intent-decoder",
label: "Intent Decoder",
desc: "Splits raw prose actions into structured intents (Player and NPC).",
},
{
key: "timedelta",
label: "TimeDelta Generator",
desc: "Calculates the duration of character actions to advance the game clock.",
},
].map((task) => (
<div key={task.key} className="mapping-card">
<div className="mapping-info">
<strong>{task.label}</strong>
<span className="text-gray" style={{ fontSize: "0.75rem", marginTop: "0.125rem" }}>{task.desc}</span>
<div
key={task.key}
className="flex flex-col justify-between gap-3 rounded-lg border border-gray-200 bg-gray-50 p-4"
>
<div className="flex flex-col gap-1 text-xs">
<strong className="text-sm text-[#111]">
{task.label}
</strong>
<span className="mt-0.5 text-gray-500">{task.desc}</span>
</div>
<select
value={mappings[task.key] || ""}
onChange={(e) => handleUpdateMapping(task.key, e.target.value)}
onChange={(e) =>
handleUpdateMapping(task.key, e.target.value)
}
className="w-full rounded border border-gray-300 bg-white px-2 py-1.5 text-xs"
>
<option value="">-- Use Active Key (Default) --</option>
<option value="">Use Default Provider</option>
{instances.map((inst) => (
<option key={inst.id} value={inst.id}>
{inst.name} ({inst.providerName}){inst.isActive ? " [Active]" : ""}
{inst.name} ({inst.providerName})
{inst.isActive ? " [Active]" : ""}
</option>
))}
</select>
@@ -334,50 +476,36 @@ export default function ConfigPage() {
</div>
</section>
<section className="config-section">
<h2>Environment Variables Default</h2>
<div className="config-row">
<span className="config-label">Default Model</span>
<span className="config-value">
<code>{config.model}</code>
</span>
</div>
<div className="config-row">
<span className="config-label">Default API Key (.env)</span>
<span
className={
config.apiKeySet
? "config-value status-ok"
: "config-value status-error"
}
>
{config.apiKeySet
? `✓ Set (${config.apiKeyPreview})`
: "✗ NOT SET"}
</span>
</div>
</section>
<section className="config-section">
<h2>Available Scenarios</h2>
<section className="mb-8 pb-6">
<h2 className="mb-3 text-lg">Available Scenarios</h2>
{config.availableScenarios.length === 0 ? (
<p className="config-hint">
No scenarios found in <code>content/demo/scenarios/</code>.
<p className="mt-3 rounded border border-amber-200 bg-amber-100 px-3 py-2 text-xs text-amber-800">
No scenarios found in{" "}
<code className="font-mono text-xs">
content/demo/scenarios/
</code>
.
</p>
) : (
<table className="scenario-table">
<table className="w-full border-collapse text-sm">
<thead>
<tr>
<th>Name</th>
<th>Path</th>
<th className="border-b-2 border-gray-200 p-2 text-left font-medium text-gray-500">
Name
</th>
<th className="border-b-2 border-gray-200 p-2 text-left font-medium text-gray-500">
Path
</th>
</tr>
</thead>
<tbody>
{config.availableScenarios.map((s) => (
<tr key={s.path}>
<td>{s.name}</td>
<td>
<code>{s.path}</code>
<td className="border-b border-gray-100 p-2">{s.name}</td>
<td className="border-b border-gray-100 p-2">
<code className="font-mono text-xs text-blue-600">
{s.path}
</code>
</td>
</tr>
))}
@@ -385,393 +513,8 @@ export default function ConfigPage() {
</table>
)}
</section>
<section className="config-section">
<h2>Engine Packages</h2>
<p className="config-hint">
All <code>@omnia/*</code> workspace packages are consumed via{" "}
<code>transpilePackages</code> in <code>next.config.ts</code>.
The native <code>better-sqlite3</code> module is externalized via{" "}
<code>serverExternalPackages</code>.
</p>
</section>
</>
</div>
)}
<style>{`
.config-page {
max-width: 800px;
margin: 0 auto;
padding: 2rem 1rem;
}
.config-page h1 {
font-size: 1.5rem;
margin-bottom: 1.5rem;
}
.config-section {
margin-bottom: 2rem;
padding-bottom: 1.5rem;
border-bottom: 1px solid #e5e7eb;
}
.config-section h2 {
font-size: 1.125rem;
margin-bottom: 0.75rem;
}
.config-row {
display: flex;
justify-content: space-between;
padding: 0.375rem 0;
border-bottom: 1px solid #f3f4f6;
}
.config-label {
color: #555;
font-size: 0.875rem;
}
.config-value {
font-size: 0.875rem;
}
.status-ok {
color: #16a34a;
}
.status-error {
color: #dc2626;
font-weight: 500;
}
.config-hint {
margin-top: 0.75rem;
padding: 0.5rem 0.75rem;
background: #fef3c7;
border: 1px solid #fde68a;
border-radius: 4px;
font-size: 0.8125rem;
color: #92400e;
}
.config-hint code {
background: rgba(0,0,0,0.06);
padding: 0.125rem 0.25rem;
border-radius: 2px;
font-size: 0.75rem;
}
.error-banner {
background: #fef2f2;
color: #b91c1c;
border: 1px solid #fca5a5;
border-radius: 4px;
padding: 0.5rem 0.75rem;
font-size: 0.875rem;
margin-bottom: 1rem;
}
.scenario-table {
width: 100%;
border-collapse: collapse;
font-size: 0.875rem;
}
.scenario-table th {
text-align: left;
padding: 0.5rem;
border-bottom: 2px solid #e5e7eb;
color: #555;
font-weight: 500;
}
.scenario-table td {
padding: 0.5rem;
border-bottom: 1px solid #f3f4f6;
}
.scenario-table code {
font-size: 0.8125rem;
color: #2563eb;
}
code {
font-family: monospace;
font-size: 0.875rem;
}
/* Pill and List Styles */
.status-pill {
display: inline-block;
padding: 0.125rem 0.5rem;
border-radius: 9999px;
font-size: 0.75rem;
font-weight: 600;
}
.status-pill.active {
background: #dcfce7;
color: #15803d;
}
.status-pill.inactive {
background: #f3f4f6;
color: #4b5563;
}
.action-buttons {
display: flex;
gap: 0.5rem;
align-items: center;
}
.btn-sm {
padding: 0.25rem 0.5rem;
font-size: 0.75rem;
background: #3b82f6;
color: #fff;
border: none;
border-radius: 4px;
cursor: pointer;
}
.btn-sm:hover {
background: #2563eb;
}
.btn-sm.delete-btn {
background: #ef4444;
}
.btn-sm.delete-btn:hover {
background: #dc2626;
}
/* Split container */
.provider-split-container {
display: grid;
grid-template-columns: 1fr;
border: 1px solid #e5e7eb;
border-radius: 12px;
overflow: hidden;
background: #fff;
margin-top: 1rem;
min-height: 400px;
}
@media (min-width: 768px) {
.provider-split-container {
grid-template-columns: 30% 70%;
}
}
/* 30% List Pane */
.provider-list-pane {
border-right: 1px solid #e5e7eb;
background: #f9fafb;
display: flex;
flex-direction: column;
}
.pane-header {
display: flex;
justify-content: space-between;
align-items: center;
padding: 1rem;
border-bottom: 1px solid #e5e7eb;
background: #f3f4f6;
}
.pane-header h3 {
margin: 0;
font-size: 0.95rem;
font-weight: 600;
color: #111;
}
.btn-add-inst {
padding: 0.375rem 0.75rem;
font-size: 0.8125rem;
font-weight: 500;
background: #10b981;
color: #fff;
border: none;
border-radius: 6px;
cursor: pointer;
transition: background 0.2s;
}
.btn-add-inst:hover {
background: #059669;
}
.pane-list {
overflow-y: auto;
flex: 1;
display: flex;
flex-direction: column;
}
.no-instances-msg {
padding: 2rem 1rem;
text-align: center;
color: #6b7280;
font-size: 0.8125rem;
}
.instance-list-item {
padding: 1rem;
border-bottom: 1px solid #e5e7eb;
cursor: pointer;
transition: background 0.15s, border-left 0.15s;
border-left: 3px solid transparent;
}
.instance-list-item:hover {
background: #f3f4f6;
}
.instance-list-item.active {
background: #eff6ff;
border-left: 3px solid #3b82f6;
}
.item-name {
font-weight: 500;
font-size: 0.875rem;
color: #111;
}
.item-meta {
display: flex;
justify-content: space-between;
align-items: center;
margin-top: 0.25rem;
font-size: 0.75rem;
color: #6b7280;
}
.active-pill {
background: #dcfce7;
color: #15803d;
font-weight: 600;
padding: 0.0625rem 0.375rem;
border-radius: 9999px;
}
/* 70% Form Pane */
.provider-form-pane {
background: #fff;
display: flex;
flex-direction: column;
}
.provider-config-form {
display: flex;
flex-direction: column;
height: 100%;
justify-content: space-between;
}
.form-scroll-content {
padding: 1.5rem;
flex: 1;
display: flex;
flex-direction: column;
gap: 1.25rem;
}
.form-scroll-content h3 {
margin: 0 0 0.5rem 0;
font-size: 1.125rem;
font-weight: 600;
color: #111;
}
.form-group {
display: flex;
flex-direction: column;
gap: 0.375rem;
}
.form-group label {
font-size: 0.8125rem;
font-weight: 500;
color: #374151;
}
.form-group input[type="text"],
.form-group input[type="password"],
.form-group select {
padding: 0.5rem 0.75rem;
font-size: 0.875rem;
border: 1px solid #d1d5db;
border-radius: 6px;
background: #fff;
outline: none;
transition: border-color 0.15s, box-shadow 0.15s;
}
.form-group input:focus,
.form-group select:focus {
border-color: #3b82f6;
box-shadow: 0 0 0 3px rgba(59, 130, 246, 0.15);
}
.checkbox-group {
flex-direction: row;
align-items: center;
gap: 0.5rem;
margin-top: 0.25rem;
}
.checkbox-group label {
cursor: pointer;
}
.checkbox-group input {
width: 1rem;
height: 1rem;
cursor: pointer;
}
/* Action bar */
.form-actions-bar {
padding: 1rem 1.5rem;
border-top: 1px solid #e5e7eb;
background: #f9fafb;
display: flex;
justify-content: space-between;
align-items: center;
}
.btn-delete-pane {
padding: 0.5rem 1rem;
font-size: 0.875rem;
font-weight: 500;
background: #ef4444;
color: #fff;
border: none;
border-radius: 6px;
cursor: pointer;
transition: background 0.2s;
}
.btn-delete-pane:hover {
background: #dc2626;
}
.btn-save-pane {
padding: 0.5rem 1.25rem;
font-size: 0.875rem;
font-weight: 500;
background: #2563eb;
color: #fff;
border: none;
border-radius: 6px;
cursor: pointer;
transition: background 0.2s;
}
.btn-save-pane:hover {
background: #1d4ed8;
}
/* Task Provider Routing Styles */
.mappings-grid {
display: grid;
grid-template-columns: 1fr;
gap: 1rem;
margin-top: 1rem;
}
@media (min-width: 768px) {
.mappings-grid {
grid-template-columns: 1fr 1fr;
}
}
.mapping-card {
border: 1px solid #e5e7eb;
background: #f9fafb;
border-radius: 8px;
padding: 1rem;
display: flex;
flex-direction: column;
justify-content: space-between;
gap: 0.75rem;
}
.mapping-info {
display: flex;
flex-direction: column;
gap: 0.25rem;
font-size: 0.8125rem;
}
.mapping-info strong {
font-size: 0.875rem;
color: #111;
}
.mapping-card select {
padding: 0.375rem 0.5rem;
font-size: 0.8125rem;
border: 1px solid #ccc;
border-radius: 4px;
background: #fff;
width: 100%;
}
.text-gray {
color: #6b7280;
}
`}</style>
</div>
);
}

View File

@@ -1,18 +1,10 @@
*,
*::before,
*::after {
box-sizing: border-box;
margin: 0;
padding: 0;
}
@tailwind base;
@tailwind components;
@tailwind utilities;
html {
font-family: system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto,
Oxygen, Ubuntu, Cantarell, sans-serif;
color: #111;
background: #fafafa;
}
body {
min-height: 100dvh;
@layer base {
html {
font-family: system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI",
Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
}
}

BIN
apps/gui/src/app/icon.png Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 919 B

View File

@@ -10,7 +10,7 @@ export const metadata = {
export default function RootLayout({ children }: { children: ReactNode }) {
return (
<html lang="en">
<body>
<body className="min-h-dvh bg-[#fafafa] text-[#111]">
<NavBar />
{children}
</body>

View File

@@ -2,62 +2,31 @@ import Link from "next/link";
export default function Home() {
return (
<main className="home">
<h1>Omnia GUI</h1>
<p className="subtitle">
<main className="mx-auto max-w-[800px] px-4 py-12">
<h1 className="mb-2 text-3xl">Omnia GUI</h1>
<p className="mb-8 text-gray-500">
Configuration and gameplay interface for the Omnia simulation engine.
</p>
<div className="home-links">
<Link href="/play" className="home-card">
<h2>Play</h2>
<p>Start a simulation and interact with NPCs</p>
<div className="flex gap-4">
<Link
href="/play"
className="block flex-1 rounded-lg border border-gray-200 p-6 text-inherit no-underline transition-[border-color,box-shadow] duration-150 hover:border-blue-600 hover:shadow-[0_2px_8px_rgba(37,99,235,0.1)]"
>
<h2 className="mb-1 text-xl">Play</h2>
<p className="text-sm text-gray-500">
Start a simulation and interact with NPCs
</p>
</Link>
<Link href="/config" className="home-card">
<h2>Config</h2>
<p>Check environment, API keys, and available scenarios</p>
<Link
href="/config"
className="block flex-1 rounded-lg border border-gray-200 p-6 text-inherit no-underline transition-[border-color,box-shadow] duration-150 hover:border-blue-600 hover:shadow-[0_2px_8px_rgba(37,99,235,0.1)]"
>
<h2 className="mb-1 text-xl">Config</h2>
<p className="text-sm text-gray-500">
Check environment, API keys, and available scenarios
</p>
</Link>
</div>
<style>{`
.home {
max-width: 800px;
margin: 0 auto;
padding: 3rem 1rem;
}
.home h1 {
font-size: 2rem;
margin-bottom: 0.5rem;
}
.subtitle {
color: #555;
margin-bottom: 2rem;
}
.home-links {
display: flex;
gap: 1rem;
}
.home-card {
flex: 1;
display: block;
padding: 1.5rem;
border: 1px solid #e5e7eb;
border-radius: 8px;
text-decoration: none;
color: inherit;
transition: border-color 0.15s, box-shadow 0.15s;
}
.home-card:hover {
border-color: #2563eb;
box-shadow: 0 2px 8px rgba(37, 99, 235, 0.1);
}
.home-card h2 {
font-size: 1.25rem;
margin-bottom: 0.25rem;
}
.home-card p {
font-size: 0.875rem;
color: #555;
}
`}</style>
</main>
);
}

View File

@@ -4,7 +4,7 @@ import path from "path";
import fs from "fs";
import { simulationManager } from "@/lib/simulation";
import type { SimSnapshot } from "@/lib/simulation";
import { ProviderManager, LLMProviderInstance } from "@omnia/llm";
import { ProviderManager, LLMProviderInstance, AVAILABLE_PROVIDERS, LLMProviderMeta } from "@omnia/llm";
function resolveScenarioPath(relative: string): string {
const cwd = process.cwd();
@@ -274,3 +274,7 @@ export async function setProviderMapping(
): Promise<void> {
ProviderManager.setMapping(task, providerInstanceId);
}
export async function getAvailableProviders(): Promise<LLMProviderMeta[]> {
return AVAILABLE_PROVIDERS;
}

View File

@@ -510,7 +510,7 @@ export function PlayView() {
)}
</select>
</div>
<button type="submit" disabled={loading}>
<button type="submit" disabled={loading || providerInstances.length === 0}>
{loading ? "Starting..." : "Start Simulation"}
</button>
</form>
@@ -534,7 +534,7 @@ export function PlayView() {
</span>
</div>
<div className="card-actions">
<button onClick={() => handleResume(s.id)} disabled={loading}>
<button onClick={() => handleResume(s.id)} disabled={loading || providerInstances.length === 0}>
Resume
</button>
<button

View File

@@ -25,7 +25,7 @@ import {
IActorProseGenerator,
buildBufferEntryForIntent,
} from "@omnia/actor";
import { GeminiProvider, ILLMProvider, MockLLMProvider, ProviderManager } from "@omnia/llm";
import { GeminiProvider, ILLMProvider, MockLLMProvider, ProviderManager, OpenRouterProvider } from "@omnia/llm";
import { ScenarioLoader } from "@omnia/scenario";
import type {
@@ -123,13 +123,6 @@ class SimulationManager {
? ProviderManager.list().find((p) => p.id === providerInstanceId)
: ProviderManager.getActive();
if (!activeInstance) {
const envKey = process.env.GOOGLE_API_KEY;
if (envKey) {
activeInstance = ProviderManager.create("Default (Env)", "google-genai", envKey);
}
}
if (!activeInstance) {
return {
id: "",
@@ -231,9 +224,12 @@ class SimulationManager {
const key = inst ? inst.apiKey : (process.env.GOOGLE_API_KEY || "");
const providerName = inst ? inst.providerName : "google-genai";
const modelName = inst ? inst.modelName : undefined;
if (providerName === "google-genai") {
return new GeminiProvider(key);
return new GeminiProvider(key, modelName);
} else if (providerName === "openrouter") {
return new OpenRouterProvider(key, modelName);
} else {
return new MockLLMProvider([]);
}
@@ -656,19 +652,15 @@ class SimulationManager {
if (!inst) {
inst = active;
}
if (!inst) {
const envKey = process.env.GOOGLE_API_KEY;
if (envKey) {
inst = ProviderManager.create("Default (Env)", "google-genai", envKey);
}
}
if (!inst) {
throw new Error(`No active LLM Provider Instance found for task "${task}". Please configure a key in Settings first.`);
}
if (inst.providerName === "google-genai") {
return new GeminiProvider(inst.apiKey);
return new GeminiProvider(inst.apiKey, inst.modelName);
} else if (inst.providerName === "openrouter") {
return new OpenRouterProvider(inst.apiKey, inst.modelName);
} else {
return new MockLLMProvider([]);
}

View File

@@ -0,0 +1,11 @@
import type { Config } from "tailwindcss";
const config: Config = {
content: ["./src/**/*.{js,ts,jsx,tsx,mdx}"],
theme: {
extend: {},
},
plugins: [],
};
export default config;

View File

@@ -8,7 +8,6 @@
"build:web": "pnpm --filter landing build",
"build:docs": "pnpm --filter docs build",
"build:gui": "pnpm --filter @omnia/gui build",
"build:all": "pnpm build && pnpm build:web && pnpm build:docs && pnpm build:gui",
"dev:web": "pnpm --filter landing dev",
"dev:docs": "pnpm --filter docs dev",
"dev:gui": "pnpm --filter @omnia/gui dev",
@@ -23,8 +22,8 @@
"test:evals": "vitest run --project evals"
},
"keywords": [],
"author": "",
"license": "ISC",
"author": "sortedcord",
"license": "MIT",
"devEngines": {
"packageManager": {
"name": "pnpm",
@@ -48,6 +47,7 @@
},
"dependencies": {
"@langchain/google-genai": "^2.2.0",
"@langchain/openrouter": "^0.4.3",
"@types/node": "^20.19.43",
"dotenv": "^17.4.2"
}

View File

@@ -2,6 +2,7 @@ import { z } from "zod";
const LLMConfigSchema = z.object({
GOOGLE_API_KEY: z.string().optional(),
OPENROUTER_API_KEY: z.string().optional(),
});
export const llmConfig = LLMConfigSchema.parse(process.env);

View File

@@ -2,4 +2,5 @@ export * from "./llm.js";
export * from "./config.js";
export * from "./providers/google-genai.js";
export * from "./providers/mock.js";
export * from "./providers/openrouter.js";
export * from "./provider-manager.js";

View File

@@ -45,3 +45,31 @@ export interface LLMProviderInstance {
isActive: boolean;
modelName?: string;
}
export interface LLMProviderMeta {
id: string;
displayName: string;
description: string;
defaultModel: string;
}
export const AVAILABLE_PROVIDERS: LLMProviderMeta[] = [
{
id: "google-genai",
displayName: "Google Gemini",
description: "Official Gemini integration using Google Gen AI SDK",
defaultModel: "gemini-2.5-flash",
},
{
id: "openrouter",
displayName: "OpenRouter",
description: "Multi-model router supporting Anthropic, OpenAI, DeepSeek, and local models",
defaultModel: "google/gemini-2.5-flash",
},
{
id: "mock",
displayName: "Mock LLM Provider",
description: "Stateless mock provider for testing and offline development",
defaultModel: "mock",
},
];

View File

@@ -3,6 +3,17 @@ import path from "path";
import fs from "fs";
import type { LLMProviderInstance } from "./llm.js";
let dbPathOverride: string | null = null;
let hasBootstrapped = false;
export function setDbPathOverride(p: string | null) {
dbPathOverride = p;
}
export function resetHasBootstrapped() {
hasBootstrapped = false;
}
function getWorkspaceRoot() {
let current = process.cwd();
while (current !== "/" && current !== path.parse(current).root) {
@@ -20,12 +31,17 @@ function getWorkspaceRoot() {
}
function getSettingsDb() {
const wsRoot = getWorkspaceRoot();
const dbDir = path.resolve(wsRoot, "data");
if (!fs.existsSync(dbDir)) {
fs.mkdirSync(dbDir, { recursive: true });
let dbPath: string;
if (dbPathOverride) {
dbPath = dbPathOverride;
} else {
const wsRoot = getWorkspaceRoot();
const dbDir = path.resolve(wsRoot, "data");
if (!fs.existsSync(dbDir)) {
fs.mkdirSync(dbDir, { recursive: true });
}
dbPath = path.join(dbDir, "settings.db");
}
const dbPath = path.join(dbDir, "settings.db");
const db = new Database(dbPath);
db.prepare(`
@@ -44,6 +60,39 @@ function getSettingsDb() {
} catch {
// ignore
}
// Auto-bootstrap environment variables if DB contains 0 instances
try {
if (!hasBootstrapped) {
const totalCount = db.prepare(`SELECT COUNT(*) as count FROM provider_instances`).get() as { count: number };
if (totalCount.count === 0) {
const googleKey = process.env.GOOGLE_API_KEY;
const openRouterKey = process.env.OPENROUTER_API_KEY;
let hasInserted = false;
if (googleKey && googleKey.trim()) {
const id = "provider-default-google";
db.prepare(`
INSERT INTO provider_instances (id, name, providerName, apiKey, isActive, modelName)
VALUES (?, ?, ?, ?, ?, ?)
`).run(id, "Gemini (Env)", "google-genai", googleKey.trim(), 1, "gemini-2.5-flash");
hasInserted = true;
}
if (openRouterKey && openRouterKey.trim()) {
const id = "provider-default-openrouter";
const isActive = hasInserted ? 0 : 1;
db.prepare(`
INSERT INTO provider_instances (id, name, providerName, apiKey, isActive, modelName)
VALUES (?, ?, ?, ?, ?, ?)
`).run(id, "OpenRouter (Env)", "openrouter", openRouterKey.trim(), isActive, "google/gemini-2.5-flash");
}
}
hasBootstrapped = true;
}
} catch {
// ignore write lock issues or other DB errors during bootstrap
}
return db;
}
@@ -153,28 +202,25 @@ export class ProviderManager {
} | undefined;
if (!row) {
// Check if there are any rows at all
const totalCount = db.prepare(`SELECT COUNT(*) as count FROM provider_instances`).get() as { count: number };
if (totalCount.count === 0) {
// Database is completely empty! Check if GOOGLE_API_KEY env is set.
const envKey = process.env.GOOGLE_API_KEY;
if (envKey && envKey.trim()) {
// Auto-bootstrap default active instance from env
const id = "provider-default-env";
db.prepare(`
INSERT INTO provider_instances (id, name, providerName, apiKey, isActive, modelName)
VALUES (?, ?, ?, ?, ?, ?)
`).run(id, "Default (Env)", "google-genai", envKey, 1, "gemini-2.5-flash");
return {
id,
name: "Default (Env)",
providerName: "google-genai",
apiKey: envKey,
isActive: true,
modelName: "gemini-2.5-flash",
};
}
// If there's no active row but some rows exist, return the first one as active, or update it
const firstRow = db.prepare(`SELECT * FROM provider_instances LIMIT 1`).get() as {
id: string;
name: string;
providerName: string;
apiKey: string;
isActive: number;
modelName?: string;
} | undefined;
if (firstRow) {
db.prepare(`UPDATE provider_instances SET isActive = 1 WHERE id = ?`).run(firstRow.id);
return {
id: firstRow.id,
name: firstRow.name,
providerName: firstRow.providerName,
apiKey: firstRow.apiKey,
isActive: true,
modelName: firstRow.modelName || undefined,
};
}
return null;
}
@@ -189,17 +235,28 @@ export class ProviderManager {
};
} catch {
// Lock or write issue fallback: return an in-memory active key if env key exists
const envKey = process.env.GOOGLE_API_KEY;
if (envKey) {
const googleKey = process.env.GOOGLE_API_KEY;
if (googleKey && googleKey.trim()) {
return {
id: "provider-default-env-fallback",
name: "Default (Env Fallback)",
name: "Gemini (Env Fallback)",
providerName: "google-genai",
apiKey: envKey,
apiKey: googleKey.trim(),
isActive: true,
modelName: "gemini-2.5-flash",
};
}
const openRouterKey = process.env.OPENROUTER_API_KEY;
if (openRouterKey && openRouterKey.trim()) {
return {
id: "provider-default-env-fallback",
name: "OpenRouter (Env Fallback)",
providerName: "openrouter",
apiKey: openRouterKey.trim(),
isActive: true,
modelName: "google/gemini-2.5-flash",
};
}
return null;
} finally {
db.close();

View File

@@ -5,6 +5,11 @@ import { llmConfig } from "../config.js";
import { ProviderManager } from "../provider-manager.js";
export class GeminiProvider implements ILLMProvider {
static readonly providerId = "google-genai";
static readonly displayName = "Google Gemini";
static readonly description = "Official Gemini integration using Google Gen AI SDK";
static readonly defaultModel = "gemini-2.5-flash";
providerName = "Gemini";
private model: ChatGoogleGenerativeAI;
lastCalls: LLMCallRecord[] = [];
@@ -15,7 +20,7 @@ export class GeminiProvider implements ILLMProvider {
if (!key) {
const active = ProviderManager.getActive();
if (active) {
if (active && active.providerName === GeminiProvider.providerId) {
key = active.apiKey;
if (!model) {
model = active.modelName;

View File

@@ -2,6 +2,11 @@ import { z } from "zod";
import { ILLMProvider, LLMRequest, LLMResponse, LLMCallRecord } from "../llm.js";
export class MockLLMProvider implements ILLMProvider {
static readonly providerId = "mock";
static readonly displayName = "Mock LLM Provider";
static readonly description = "Stateless mock provider for testing and offline development";
static readonly defaultModel = "mock";
providerName = "mock";
private callCount = 0;
lastCalls: LLMCallRecord[] = [];

View File

@@ -0,0 +1,80 @@
import { z } from "zod";
import { ChatOpenRouter } from "@langchain/openrouter";
import { ILLMProvider, LLMRequest, LLMResponse, LLMCallRecord } from "../llm.js";
import { llmConfig } from "../config.js";
import { ProviderManager } from "../provider-manager.js";
export class OpenRouterProvider implements ILLMProvider {
static readonly providerId = "openrouter";
static readonly displayName = "OpenRouter";
static readonly description = "Multi-model router supporting Anthropic, OpenAI, DeepSeek, and local models";
static readonly defaultModel = "google/gemini-2.5-flash";
providerName = "OpenRouter";
private model: ChatOpenRouter;
lastCalls: LLMCallRecord[] = [];
constructor(apiKey?: string, modelName?: string) {
let key = apiKey;
let model = modelName;
if (!key) {
const active = ProviderManager.getActive();
if (active && active.providerName === OpenRouterProvider.providerId) {
key = active.apiKey;
if (!model) {
model = active.modelName;
}
}
}
if (!key) {
key = llmConfig.OPENROUTER_API_KEY;
}
if (!key) {
throw new Error("OPENROUTER_API_KEY is required to initialize OpenRouterProvider");
}
this.model = new ChatOpenRouter({
apiKey: key,
model: model || "google/gemini-2.5-flash",
});
}
async generateStructuredResponse<T extends z.ZodTypeAny>(
request: LLMRequest<T>,
): Promise<LLMResponse<z.infer<T>>> {
const structuredModel = this.model.withStructuredOutput(request.schema, { includeRaw: true });
const result = (await structuredModel.invoke([
{ role: "system", content: request.systemPrompt },
{ role: "user", content: request.userContext },
])) as unknown as {
parsed?: z.infer<T>;
raw?: {
usage_metadata?: {
input_tokens?: number;
output_tokens?: number;
total_tokens?: number;
};
};
};
const parsed = result?.parsed;
const raw = result?.raw;
const usage = raw?.usage_metadata ? {
inputTokens: raw.usage_metadata.input_tokens || 0,
outputTokens: raw.usage_metadata.output_tokens || 0,
totalTokens: raw.usage_metadata.total_tokens || 0,
} : undefined;
this.lastCalls.push({
systemPrompt: request.systemPrompt,
userContext: request.userContext,
usage,
});
return { success: true, data: parsed, usage };
}
}

View File

@@ -0,0 +1,107 @@
import { describe, test, expect, vi } from "vitest";
import { z } from "zod";
import { OpenRouterProvider } from "../src/providers/openrouter.js";
import { llmConfig } from "../src/config.js";
// Mock the ChatOpenRouter class
vi.mock("@langchain/openrouter", () => {
return {
ChatOpenRouter: class {
config: unknown;
constructor(config: unknown) {
this.config = config;
}
withStructuredOutput = vi.fn().mockImplementation(() => {
return {
invoke: vi.fn().mockImplementation(async () => {
// Return a mock output that matches the includeRaw: true structure
return {
parsed: {
name: "mocked response",
success: true,
},
raw: {
usage_metadata: {
input_tokens: 10,
output_tokens: 5,
total_tokens: 15,
},
},
};
}),
};
});
},
};
});
describe("OpenRouterProvider Unit Tests (Tier 1)", () => {
test("initializes successfully with a provided apiKey", () => {
const provider = new OpenRouterProvider("dummy-key");
expect(provider.providerName).toBe("OpenRouter");
});
test("initializes successfully with apiKey from config", () => {
// Save current config
const originalKey = llmConfig.OPENROUTER_API_KEY;
llmConfig.OPENROUTER_API_KEY = "env-dummy-key";
try {
const provider = new OpenRouterProvider();
expect(provider.providerName).toBe("OpenRouter");
} finally {
llmConfig.OPENROUTER_API_KEY = originalKey;
}
});
test("throws error if no API key is provided or in config", () => {
// Save current config
const originalKey = llmConfig.OPENROUTER_API_KEY;
llmConfig.OPENROUTER_API_KEY = undefined;
try {
expect(() => new OpenRouterProvider()).toThrow(
"OPENROUTER_API_KEY is required to initialize OpenRouterProvider"
);
} finally {
llmConfig.OPENROUTER_API_KEY = originalKey;
}
});
test("generateStructuredResponse invokes the model with structured output, records usage and updates lastCalls", async () => {
const provider = new OpenRouterProvider("dummy-key");
const TestSchema = z.object({
name: z.string(),
success: z.boolean(),
});
const response = await provider.generateStructuredResponse({
systemPrompt: "system prompt",
userContext: "user context",
schema: TestSchema,
});
expect(response.success).toBe(true);
expect(response.data).toEqual({
name: "mocked response",
success: true,
});
expect(response.usage).toEqual({
inputTokens: 10,
outputTokens: 5,
totalTokens: 15,
});
expect(provider.lastCalls.length).toBe(1);
expect(provider.lastCalls[0]).toEqual({
systemPrompt: "system prompt",
userContext: "user context",
usage: {
inputTokens: 10,
outputTokens: 5,
totalTokens: 15,
},
});
});
});

View File

@@ -0,0 +1,91 @@
import { describe, test, expect, beforeEach, afterEach } from "vitest";
import fs from "fs";
import path from "path";
import { ProviderManager, setDbPathOverride, resetHasBootstrapped } from "../src/index.js";
describe("ProviderManager Bootstrapping & CRUD Unit Tests", () => {
let tempDbPath: string;
let originalGoogle: string | undefined;
let originalOpenRouter: string | undefined;
beforeEach(() => {
originalGoogle = process.env.GOOGLE_API_KEY;
originalOpenRouter = process.env.OPENROUTER_API_KEY;
delete process.env.GOOGLE_API_KEY;
delete process.env.OPENROUTER_API_KEY;
resetHasBootstrapped();
// Generate a unique temp database path for this test run
tempDbPath = path.resolve(process.cwd(), `test-settings-${Date.now()}-${Math.random().toString(36).substring(2)}.db`);
setDbPathOverride(tempDbPath);
});
afterEach(() => {
setDbPathOverride(null);
if (fs.existsSync(tempDbPath)) {
try {
fs.unlinkSync(tempDbPath);
} catch {
// ignore
}
}
if (originalGoogle !== undefined) {
process.env.GOOGLE_API_KEY = originalGoogle;
} else {
delete process.env.GOOGLE_API_KEY;
}
if (originalOpenRouter !== undefined) {
process.env.OPENROUTER_API_KEY = originalOpenRouter;
} else {
delete process.env.OPENROUTER_API_KEY;
}
});
test("auto-bootstraps Gemini and OpenRouter when database is empty and environment variables are present", () => {
process.env.GOOGLE_API_KEY = "mock-google-key-123";
process.env.OPENROUTER_API_KEY = "mock-openrouter-key-456";
const list = ProviderManager.list();
expect(list.length).toBe(2);
const gemini = list.find((p) => p.providerName === "google-genai");
expect(gemini).toBeDefined();
expect(gemini?.name).toBe("Gemini (Env)");
expect(gemini?.apiKey).toBe("mock-google-key-123");
expect(gemini?.modelName).toBe("gemini-2.5-flash");
expect(gemini?.isActive).toBe(true); // first inserted is active
const openrouter = list.find((p) => p.providerName === "openrouter");
expect(openrouter).toBeDefined();
expect(openrouter?.name).toBe("OpenRouter (Env)");
expect(openrouter?.apiKey).toBe("mock-openrouter-key-456");
expect(openrouter?.modelName).toBe("google/gemini-2.5-flash");
expect(openrouter?.isActive).toBe(false); // second inserted is inactive
});
test("treats bootstrapped instances as normal provider instances (editable and deletable)", () => {
process.env.GOOGLE_API_KEY = "mock-google-key-123";
// Trigger bootstrap
const list = ProviderManager.list();
expect(list.length).toBe(1);
const bootstrapped = list[0];
expect(bootstrapped.name).toBe("Gemini (Env)");
expect(bootstrapped.isActive).toBe(true);
// Edit name and key
ProviderManager.update(bootstrapped.id, "My Gemini Key", "google-genai", "new-secret-key", "gemini-2.5-pro");
const listAfterUpdate = ProviderManager.list();
expect(listAfterUpdate.length).toBe(1);
expect(listAfterUpdate[0].name).toBe("My Gemini Key");
expect(listAfterUpdate[0].apiKey).toBe("new-secret-key");
expect(listAfterUpdate[0].modelName).toBe("gemini-2.5-pro");
// Delete instance
ProviderManager.delete(bootstrapped.id);
const listAfterDelete = ProviderManager.list();
expect(listAfterDelete.length).toBe(0);
});
});

1680
pnpm-lock.yaml generated

File diff suppressed because it is too large Load Diff

View File

@@ -8,6 +8,7 @@ allowBuilds:
esbuild: true
sharp: true
unrs-resolver: true
workerd: true
minimumReleaseAgeExclude:
- '@astrojs/telemetry@3.3.3'
- astro@7.0.7

View File

@@ -5,10 +5,12 @@ import mermaid from "astro-mermaid";
export default defineConfig({
site: "https://omnia.omniasimulation.com",
base: "/docs",
outDir: "./dist/docs",
integrations: [
mermaid(),
starlight({
title: "Omnia Docs",
favicon: "/favicon.png",
logo: {
src: "./src/assets/img/logo.png",
replacesTitle: true,

View File

@@ -7,11 +7,17 @@
"dev": "astro dev",
"start": "astro dev",
"build": "astro build",
"preview": "astro preview"
"preview": "astro preview",
"preview:wrangler": "astro build && wrangler dev"
},
"dependencies": {
"@astrojs/starlight": "^0.41.3",
"astro": "^7.0.7",
"astro-mermaid": "^2.1.0",
"mermaid": "^11.16.0",
"sharp": "^0.33.5"
},
"devDependencies": {
"wrangler": "^4.107.1"
}
}

BIN
web/docs/public/favicon.png Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 919 B

Binary file not shown.

After

Width:  |  Height:  |  Size: 184 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 919 B

Binary file not shown.

After

Width:  |  Height:  |  Size: 81 KiB

View File

@@ -21,9 +21,10 @@ export interface ILLMProvider {
}
```
The codebase provides two primary implementations:
The codebase provides three primary implementations:
1. **`GeminiProvider`:** The production provider utilizing Google's Gemini Models via the `@langchain/google-genai` SDK.
2. **`MockLLMProvider`:** A stateless, pre-programmed mock provider used for fast, deterministic unit testing and local integration tests.
2. **`OpenRouterProvider`:** The production provider utilizing OpenRouter via the `@langchain/openrouter` SDK, allowing routing through various third-party and local models.
3. **`MockLLMProvider`:** A stateless, pre-programmed mock provider used for fast, deterministic unit testing and local integration tests.
---
@@ -43,10 +44,10 @@ export interface LLMProviderInstance {
```
Users can register multiple provider instances in the **Configuration Page** under the GUI. Each instance is given:
* A friendly, human-readable name (e.g., `"Gemini Production Key"`, `"Experimental Gemini Pro"`).
* A provider type (e.g., `google-genai`, `mock`).
* A friendly, human-readable name (e.g., `"Gemini Production Key"`, `"OpenRouter Claude Key"`).
* A provider type (e.g., `google-genai`, `openrouter`, `mock`).
* An API key credential.
* A custom target model name (e.g., `gemini-2.5-flash` or `gemini-2.5-pro`).
* A custom target model name (e.g., `gemini-2.5-flash`, `anthropic/claude-3-5-sonnet`, or local model paths).
* An **Active** status flag (one key is marked as globally active).
Configurations are stored globally in `data/settings.db` (separated from specific simulation run databases like `data/sim-*.db` to keep key storage and audit logs isolated).
@@ -59,10 +60,10 @@ During a simulation run, the engine executes four distinct LLM operations. To op
| Task Name | Key ID | Description | Default Model |
| :--- | :--- | :--- | :--- |
| **Actor Prose Generation** | `actor-prose` | Generates roleplay and narrative behavioral prose for Non-Player Characters. | `gemini-2.5-flash` |
| **LLM Validator** | `llm-validator` | Arbitrates and validates proposed actions against the world state rules and constraints. | `gemini-2.5-flash` |
| **Intent Decoder** | `intent-decoder` | Parses and splits free-text actions/prose into structured intent sequences. | `gemini-2.5-flash` |
| **TimeDelta Generator** | `timedelta` | Calculates the duration of character actions to advance the game clock. | `gemini-2.5-flash` |
| **Actor Prose Generation** | `actor-prose` | Generates roleplay and narrative behavioral prose for Non-Player Characters. | `gemini-2.5-flash` / `google/gemini-2.5-flash` |
| **LLM Validator** | `llm-validator` | Arbitrates and validates proposed actions against the world state rules and constraints. | `gemini-2.5-flash` / `google/gemini-2.5-flash` |
| **Intent Decoder** | `intent-decoder` | Parses and splits free-text actions/prose into structured intent sequences. | `gemini-2.5-flash` / `google/gemini-2.5-flash` |
| **TimeDelta Generator** | `timedelta` | Calculates the duration of character actions to advance the game clock. | `gemini-2.5-flash` / `google/gemini-2.5-flash` |
If no specific provider instance is mapped to a task, the task automatically routes to the globally marked **Active** provider instance.
@@ -72,9 +73,10 @@ If no specific provider instance is mapped to a task, the task automatically rou
To maintain backwards-compatibility and support headless runs, live evaluation suites, and automated unit tests without requiring database pre-configuration, the config manager supports **self-bootstrapping**:
1. When the provider manager queries the active key instance, if `data/settings.db` contains **0 registered keys**, it checks the process environment for `GOOGLE_API_KEY`.
2. If `process.env.GOOGLE_API_KEY` is present, it automatically creates, saves, and activates a default provider instance (`Default (Env)`) in `settings.db`.
3. If database write locks occur (e.g., during high-concurrency Vitest test suites), the system seamlessly returns a temporary in-memory `LLMProviderInstance` to keep execution fluent and error-free.
1. When any database connection is initialized via the provider manager, if `data/settings.db` contains **0 registered keys**, it checks the process environment for `GOOGLE_API_KEY` and `OPENROUTER_API_KEY`.
2. If `process.env.GOOGLE_API_KEY` is present, it automatically creates, saves, and activates a default provider instance (`Gemini (Env)`) in `settings.db`.
3. If `process.env.OPENROUTER_API_KEY` is present, it automatically creates and saves a default provider instance (`OpenRouter (Env)`) in `settings.db`.
4. If database write locks occur (e.g., during high-concurrency Vitest test suites), the system seamlessly returns a temporary in-memory `LLMProviderInstance` (`Gemini (Env Fallback)` or `OpenRouter (Env Fallback)`) to keep execution fluent and error-free.
---

14
web/docs/wrangler.jsonc Normal file
View File

@@ -0,0 +1,14 @@
{
"$schema": "node_modules/wrangler/config-schema.json",
"name": "omnia-docs",
"compatibility_date": "2026-07-09",
"assets": {
"directory": "./dist"
},
"routes": [
{
"pattern": "omnia.adityagupta.dev/docs*",
"zone_name": "adityagupta.dev"
}
]
}