Search

Marginal SDK for TypeScript/JavaScript — track AI cost per customer and feature. Wrap your OpenAI / Anthropic / Gemini client or register the Vercel AI SDK integration; events are priced server-side.

On-device Expo OCR for Afghan identity documents (ML Kit/Vision + PP-OCR, optional B&W preprocess)

Assistant relay protocol: the session handshake, turn events, and tool-call wire the hosted Assistant's split loop speaks

Pi extension: /usage shows subscription quota meters for Claude, Codex, Copilot, Kimi, Grok, GLM, and MiniMax accounts, plus DeepSeek balances.

Install a file-backed, spec-driven AI development workflow into a scaffolded app.

Thin gRPC client for the Ultipa graph database (ultipa-gqldb) — the Ultipa driver behind Tina4's graph layer

Node.js screenshot capture for Storybook Test Runner - Provides stable screenshot functionality using Playwright for visual regression testing

Public contracts, validation, and testing utilities for OpsRabbit plugins

Framework-agnostic PDF annotation engine — render a PDF and let users draw, highlight, and write on it.

Add CollabHub to an existing React app or create a minimal example

神治（Thearchy）— deterministic multi-agent quality governance for coding agents.

Augmented Context — a context engineering harness for AI coding agents. Code graphs, hybrid retrieval, repository compression, multi-signal ranking, and token-aware context assembly.

Subscription quota in the pi status bar for Ark (Volcengine), Kimi For Coding, and Ollama Cloud

Jugnu — a terminal-based AI coding assistant with an Ink TUI.

Unified LLM client — single interface for OpenAI, Anthropic, DeepSeek, Ollama

DeepSeek Harness API balance and local token usage plugin

DSH plugin: electrical & electronics problem solving

dsh plugin: back the native web_search / web_fetch tools with your self-hosted SearXNG instance - keyless, private, no third-party search vendor.

灵枢（Lingshu·líng shū）DeepSeek Harness 插件：完整大脑——长期记忆/知识飞轮/自我认知/递归反思接入 DSH，对话自动沉淀进记忆库

A middle permission tier for DeepSeek Harness between workspace-write and danger-full-access: auto-approves harmless commands and operations targeting configured trusted areas, beyond the current workspace.