x12i

AI model catalogs (open-assets JSON), cost calculation, and LLM utilities

Download a URL with a real (headless) browser and convert it to clean Markdown, producing the FetchResult that @x12i/core parsers consume.

Bundled model catalog reads: identity, lane, and price, without profile resolution

Offline, JSON-backed lookup of AWS AI service and foundation-model regional availability (Bedrock access modes included). Unofficial / community-maintained — not an AWS product.

Offline, JSON-backed lookup of Azure AI / Azure OpenAI regional availability. Unofficial / community-maintained — not a Microsoft product.

Catalox object storage (Cloudflare R2 / S3-compatible)

ExecutableGraphPlan format: plan schema validation and canonicalization.

Graphenix core tier: format 2.1.0 JSON schema, validation, and CRUD helpers.

Manage Memorix entity, list, and item descriptors in Catalox

Memorix unified durable pipeline runtime (memorix-pipeline/v2)

Shared Mongo operational repositories for org-scoped artifacts, receipts, DLQ, leases, checkpoints, and schedule leases

Shared demo domain, descriptors, seed records, and stage helpers for Memorix Data & Pipeline Simulators

Memorix data plane: scope/routing (5A), V2 writer (5B), reader (5C), abstract reverse-write (5D), relationship store/read (5E).

Memorix Connector SDK — defineConnector, workflow, checkpoint-aware collectStream, host/remote/testing surfaces, and conformance utilities.

First-party KnowX agent pack: folder sources, content model, extract/associate services, pipelines, control plane.

LLM-backed worox-graph composer: create, modify, explain, suggestConceptObjective, reviewConcept, catalog resolution/creation, scoping need match and scoping map creation plans, plus optional orchestrator — via OpenRouter and aifunctions-js

FuncX: LLM function runtime with Catalox-backed content catalog.

Unified router for all LLM provider implementations

Inline variant template rendering with Handlebars

Extensible Multi-Format AI-Powered Markdown to JSON Transformer with support for any markdown variant through custom parsers, intelligent schema validation, auto-fixing, table parsing, and optional persistent machine learning

Git-like versioning engine for MongoDB metadata (metagit).

AI Gateway - Unified interface for LLM provider routing and management

Graphenix graph description format: JSON schema, validation, and CRUD helpers.

Platform infrastructure for reusable, interoperable catalogs across apps and data sources.

Resolve AI profile names into concrete model choices with bundled defaults and remote JSON refresh

OpenAI provider implementation for LLM providers

Scenario-based runtime optimization engine for AI workloads; first scenario ai.max_tokens.v1 (max completion budget), integrated with Activix

Descriptor-driven write layer for Memorix entity/event records and content objects

Configuration normalizer for Node.js - Zero dependencies, TypeScript-first, production-ready

HTTP client for the Authix Token Infrastructure API

SDK-first Provider AI Interface for router-adapter architecture. Supports sync, streaming, and batch operations with lossless raw vendor responses.

CLI for automating npm and git workflows across one or more packages

GraphExecutionTrace format: append-only events, validation, and trace helpers.

Generic two-phase activity lifecycle tracker built on @x12i/xronox-store

HTTP client for Catalox API

Standardized logging gateway for Node.js with PII sanitization, correlation trails, Shadow Logging for test/debug capture, scoping with text filters, story output, troubleshooting integration, and multi-format output (JSON/YAML/text)

AI provider adapter for xAI Grok, implementing @athenices/ai-provider-interface

TypeScript runtime for OpenRouter models, server tools, user tools, citations, patches, and normalized responses.

Unified AI router runtime. Provider-agnostic entrypoint with OpenRouter protocol; OpenRouter plus local llama-cpp and transformersjs providers.

KnowX living knowledge graph Web Component (static Explorer + live memorix-service)

Folder-based JSON-returning AI assistants — callable as simple functions with a router and test()

Typed HTTP client for the Connectix service: discovery routes, canonical execute, and Memorix-shaped pipeline calls.

Git-like versioning engine for MongoDB metadata (MaGit).

Metadata-driven generic source connector runtime for Memorix

Metadata-driven generic target connector runtime for Memorix

Client SDK for Credorix token leases and authenticated requests.

Published generic Memorix connector runtime (source + target tools)

Direction-neutral foundation for Memorix generic connectors

Durable multi-step session runtime that hosts MGC vendor pulls and pure stages over Explorer

CLI tool to generate MP4 videos by matching image (.png/.jpg/.jpeg) and audio (.wav/.mp3) pairs, plus compose multi-part timelines with optional audio mix and per-cut transitions.

Generic org-isolated lexical, semantic, and hybrid retrieval companion for Memorix

Nexum — Memorix analyze worker + explore SDK. Deterministic knowledge-graph drafts; optional Mongo ingest via Python REST; semantic via assistant-functions.

Browser HTTP client for map-store /map-store/v1 (Memorix estate via map-store only)

Process Memorix records through a skill — select by OT/CT filter, run skill with Scoper MCP, markdown→JSON, write output CT

The @x12i Edge Data Stack: dual-layer hybrid docstore and graph database for edge runtimes

Skill runner — load an agent processing and auditing *.md, render with Rendrix, run via @x12i/ai-dispatcher

Canonical AI call record schema (JSON Schema v1.0), TypeScript types, metadata rules and model aliases for xtivix

HTTP client for the AI Dispatcher API. Call run and compile over the network.