AI2Web (ai2w) core - types, capability model, manifest builder, validation & AI Readiness scoring.

The Archstone compiler — validates a capability model and lowers it to a target-agnostic intermediate representation that every emitter consumes.

Neutral ToolpathIR and capability model for the Chestnut Labs G-code Preview toolpath stack.

Neutral capability model and governed agent runtime for oRPC procedures. Independent community project, not affiliated with or endorsed by the oRPC maintainers.

Athos runtime contracts and capability model

Generic pluggable message broker abstraction. Domain-agnostic pub/sub with capability model. In-process default, external adapters (Kafka, RabbitMQ, …) implement the same Broker interface. Zero external npm dependencies.

Generic pluggable notification-channel abstraction. Domain-agnostic send-to-recipient with a capability model and a zero-dependency template renderer. In-process default, external adapters (Webhook, Slack, Email, SMS, …) implement the same Channel interfa

Schema.org vocabulary plus the .org.ai superset (Agent, Tool, Capability, Model, …) as TypeScript types.

This library provides a wrapper around utilities to prepare a model process to run

Public SDK for building Verql plugins — types, helpers, and the capability model third-party plugin authors code against.

MCP server for AIProx — the open agent registry. Discover and hire 24 verified autonomous agents by capability. Model selection passthrough: specify any of 19 LightningProx models per agent call. Supports Bitcoin Lightning, Solana USDC, and Base x402.

A package that provides 'Sign in with Apple' capability for Expo and React Native apps.

JavaScript/TypeScript-native RPC library with Promise Pipelining

Fine-tune cheap open-source LLMs (GLM, Qwen, DeepSeek) on your AI coding agent's successful runs with LoRA (SFT + DPO) so your model cascade escalates to expensive frontier models (GPT, Claude) less often — cutting cost-per-resolved. Turns run history int

Workspace-aware Databricks Model Serving selection.

AI red-teaming for the AI agents & LLM apps you own: stress-test them with adversarial models to find security failures (prompt injection, tool misuse / excessive agency, data leakage, jailbreaks, denial-of-wallet), auto-patch (blue team), retest, and get

Capability model and policy enforcement for wisp framework

Provider-neutral intelligence infrastructure for NestJS – language, embeddings, media, reranking, realtime, agents, tools, diagnostics, files, and skills.

Cryptographic trust infrastructure for MCP servers — content-addressed tool identity, signed manifests, TOFU pinning, capability descriptors, a two-gate invocation policy, signed+chained receipts, and publisher key rotation. Also the "npm audit" health-ch

NubemClaw v3 — Hermes-ported high-value tools (file, terminal, code_execution, browser, web, memory, delegate, mcp, image, tts, vision, send_message, todo, session_search). Conceptual ports adapted to v3 sandbox + capability model. See PORTING_NOTES.md fo