SuperAPI AI Directory
Partner with us中文AI newsFavorites

AI developer platforms

3,655 results · Page 50 of 102

agency-swarm

AI developer platforms

Agency Swarm is an open-source framework for building multi-agent applications, built on and extending the OpenAI Agents SDK, providing specialized capabilities for creating, orchestrating, and managing collaborative AI agent swarms. It continues the original vision of Arsenii Shatokhin (also known as VRSEN), thinking about automation in terms of real-world organizational structures, making it intuitive for both agents and users to understand. The framework provides customizable agent roles, full control over prompts and instructions, type-safe tools based on Pydantic, agent communication implemented through the send_message tool and constrained by explicit directed communication flows, flexible session state persistence, and multi-agent orchestration capabilities, and is designed for production-grade reliability. The v1.x version targets the OpenAI Agents SDK and the Responses API. Model backends support both native OpenAI (such as the GPT-5 series, GPT-4o, etc.) and, through LiteLLM routing, access to Anthropic Claude, Google Gemini, Grok (xAI), Azure OpenAI, OpenRouter, and more. It supports macOS, Linux, and Windows. The project is open source and uses the MIT license.

Read more

agent-express

AI developer platforms

agent-express is a fast, minimal AI agent framework written in TypeScript, adopting an Express.js-like middleware pattern. It uses only three core concepts—Agent, Session, and Middleware—replacing the numerous concepts in other frameworks with a unified (ctx, next) interface. The framework provides 5 layers of onion-style middleware hooks (agent, session, turn, model, tool), and has built-in guard capabilities such as budget caps, input/output validation, timeouts, iteration limits, and human approval, and supports more than 12 model providers, complexity-based model routing, context window compaction, MCP integration, and SSE streaming output.

Read more

agent-sdk-go

AI developer platforms

agent-sdk-go is an open-source SDK for building persistent AI agents in the Go language. Its core feature is that it allows an agent's state, tool calls, and execution loop to recover after a process crash, thereby "continuing from where it left off." By default it runs in-process, providing crash recovery without additional infrastructure; when distributed, multi-process execution is needed, it can switch to Temporal or Restate. Each core component of the SDK is a pluggable interface, avoiding vendor lock-in. It supports LLM providers such as OpenAI, Anthropic, Gemini, DeepSeek, Ollama, and more; supports built-in and custom tools, MCP servers, the A2A protocol, sub-agents, human approval, multi-turn conversation history, memory and RAG, streaming output and AG-UI, extended reasoning, token usage statistics, budget control, lifecycle hooks, error control, execution configuration, distributed execution, and OpenTelemetry observability. It requires Go 1.26+ and is open-sourced under the Apache License 2.0. This project is an independent community library and is not affiliated with Temporal Technologies or Restate.

Read more

agent-skills

AI developer platforms

Agent Skills is the reference implementation of ORCA (Open Cognitive Runtime Architecture), positioned as a "runtime for executable cognition." It extracts repeatable agent reasoning logic from prompts and turns it into executable, reusable, testable, and observable Skills, allowing agents to execute directly when possible rather than relying only on prompts. The project's core concepts include: Skills (reusable cognitive workflows declared as DAGs), Capabilities (backend-agnostic contracts with typed inputs and outputs), Bindings (connecting contracts to execution backends such as PythonCall, OpenAPI, MCP, and OpenRPC), and Runtime (an execution layer with DAG scheduling, policy and security gating, CognitiveState, and traceability). Running locally requires no API key and provides a deterministic Python baseline for offline development and testing; the first run may take 30-60 seconds. The repository also provides an HTTP API, an MCP server, framework adapters (LangChain, CrewAI, AutoGen, Semantic Kernel), and integration methods such as native tool definitions for Anthropic, OpenAI, and Gemini, and is open-sourced under the Apache 2.0 license.

Read more

AgentDock

AI developer platforms

AgentDock is a TypeScript-based AI agent development framework for building AI agents capable of delivering complex tasks with configurable determinism. It consists of AgentDock Core (an open-source, backend-first, framework-agnostic, vendor-independent agent core framework) and an open-source Next.js-based client reference implementation. The framework emphasizes simplicity, extensibility, and configurable determinism, making it suitable for building reliable and predictable AI systems that can operate with minimal supervision. It requires Node.js ≥ 20.11.0, pnpm ≥ 9.15.0, and an API key from at least one LLM provider. The framework repository has been retired; the prompt and documentation generator has been migrated to AgentDock/prompt-library, and the book has been migrated to AgentDock/ai-agents-book.

Read more

AgentFlow

AI developer platforms

AgentFlow is a trainable, tool-integrated agent framework developed by lupantech, designed to overcome the scalability and generalization limitations of tool-augmented reasoning methods. It adopts a modular design comprising four specialized modules: Planner, Executor, Verifier, and Generator, which work collaboratively across multiple rounds of interaction through evolving memory and integrated tools. The framework uses Flow-based Group Refined Policy Optimization (Flow-GRPO) to optimize the Planner online, performing excellently on 10 benchmarks spanning search, agentic reasoning, mathematics, and science, with a 7B model even surpassing GPT-4o with around 200B parameters. The framework supports multi-tool integration (such as Python coding, Google search, Wikipedia search, etc.) and provides inference and training pipelines. The project has been accepted by ICLR 2026 and the NeurIPS 2025 workshop, and ranked second in HuggingFace Daily Paper.

Read more

Agentic-RAG-R1

AI developer platforms

Agentic-RAG-R1 (repository name Harness-RL) is a reinforcement learning-based agentic RAG framework, and the official implementation of the paper accepted at PCC 2026. It proposes training language model agents within a central-agent multi-agent execution framework (harness), without flattening the execution process into a single token sequence, but instead constructing black-box trajectories from interface calls and session prefix trees, and employing Conflict-Aware Policy Optimization (CAPO) to route the gradients of action tokens and argument tokens separately to their corresponding parameter subspaces. The framework supports central-only agent training and joint multi-agent training, integrates the production-grade agent framework StackPlanner and the slime distributed training stack, and targets Linux x86_64 + NVIDIA GPU environments.

Read more

AgentLab

AI developer platforms

AgentLab is an open-source framework for developing, testing, and benchmarking Web Agents, focusing on scalability and reproducibility. Provided by ServiceNow and built on BrowserGym, the framework supports large-scale parallel experiments and multiple benchmarks. AgentLab is primarily for research purposes and is not a consumer-facing product, so risks should be noted when using it. It provides foundational modules for building agents, unified LLM API support (OpenRouter, OpenAI, Azure, or self-hosted TGI), a recommended way to run benchmarks such as WebArena, multiple reproducibility features, and a unified leaderboard.

Read more

agentlego

AI developer platforms

AgentLego is a tool API library open-sourced by the InternLM team, used to extend and enhance intelligent agents based on large language models (LLMs). It provides a rich set of multimodal tools covering capabilities such as visual perception, image generation and editing, speech processing, and vision-language reasoning, and supports flexible tool interfaces, allowing users to extend custom tools with any type of parameters and outputs. AgentLego can be easily integrated into mainstream agent frameworks such as LangChain, Transformers Agents, and Lagent, and also supports tool serviceization and remote access, making it convenient for use in scenarios that require heavy machine learning models (such as ViT) or special environments (such as GPU, CUDA).

Read more

agentops

AI developer platforms

AgentOps is a Python SDK for AI agent monitoring, LLM cost tracking, and benchmarking, supporting the development process from prototype to production. It automatically captures analytics data for all LLM calls, and sessions can be viewed in the AgentOps dashboard. AgentOps provides a first-class developer experience, adding powerful observability to agents, tools, and functions with minimal code (added line by line), supporting input/output recording, exception handling, async/await functions, generator functions, and custom attributes and names. It integrates with most LLM and agent frameworks, including CrewAI, Agno, OpenAI Agents SDK, Langchain, Autogen, AG2, CamelAI, Cohere, Anthropic, Mistral, LiteLLM, LlamaIndex, Llama Stack, SwarmZero, and more. The AgentOps application itself is open source under the MIT license.

Read more