Search

An LLM development tool like requests.

LLM output evaluation metrics for Node.js/TypeScript

Policy-driven guardrails middleware for LLM requests (PII masking, prompt-injection detection)

High-performance OpenAI-compatible proxy reverse-engineering ChatGPT Web with zero-auth Gemini Flash fallback

类型/接口/能力声明包（只导出 `type` 与 `*Capabilities`，不含运行时逻辑）

Governance + orchestration framework wrapping the Anthropic Claude API for enterprise deployments

The OpenAI-compatible chat-completions provider as a Stratus plugin: registers `openai-compatible`, selectable by a soul, credentialed per agent through the plugin seam

Voice-driven real-time interview coach: records audio, transcribes with Whisper, answers via Claude/GPT/DeepSeek/Gemini with meeting plans, briefing research, and screen capture.

TypeScript client for the Vovix LLM API: chat, structured output and tool-calling agents on Amazon Bedrock Mantle. Zero dependencies.

No description available

Partition documents into LLM-ready elements. The call path of unstructured, with zero required dependencies.

LLM API security auditor — prompt injection, jailbreak, ASCII smuggling, data exfiltration, OWASP LLM Top 10 mapping

Git history analyzer with LLM-powered narrative generation

Per-model LLM call rate limiter for DeepSeek Harness with queue support

Universal OpenAI-compatible sidecar proxy with a live ragfaith faithfulness cascade

A Claude Code skill that generates images through OpenAI, Gemini, Higgsfield, FLUX, or any OpenAI-compatible endpoint. One CLI, no dependencies.

Agent-native shopping for extreme value: verifiable same-product price evidence, checkout, orders, delivery, and after-sales.

The Open Registry for AI Engineering Presets.

我的 Claude Code skills 集合，用于提升开发效率

Jina v5 text embeddings via ONNX Runtime (Rust core, PyO3 bindings)