Agent frameworks & infrastructure
Agent frameworks & SDKs
12,596 agents · most starred first
Agent frameworks & SDKs, page 126 of 126
Default retry policy plugin for Agent Forge (retryable-error classification and exponential backoff)
Agents as files, harness included: zero-dependency agent runtime with budgets, permissions, durable resume, multi-agent patterns, MCP and OpenTelemetry.
Embeddable TypeScript agent runtime with policy, human-in-the-loop approvals, checkpoints, and tool calling.
Generate a skein-js LangGraph application in an Nx workspace.
CrewAI Python integration for Agent-DID identities, compatible tools, callbacks and secure opt-in operations.
LangChain Python integration for Agent-DID identities, context injection and secure tool exposure.
Microsoft Agent Framework Python integration for Agent-DID identities, native tools and sanitized observability.
Semantic Kernel Python integration for Agent-DID identities, tools, context injection and secure opt-in operations.
Open-source Python harness for building reliable AI agents.
A local-first framework for agentic software development workflows.
AgenticWF agent client: the Python AgentClient and the `agenticwf agent` command
AgenticWF client: the `agenticwf` command (Logto sign-in, tokens, API calls; `kb` and `agent` commands from their client packages)
AgenticWF knowledge service client: the Python KBClient and the `agenticwf kb` command
AgentMall SDK placeholder — https://getagentmall.com
Azure-backed providers for the Agent Skills format (https://agentskills.io)
Git-backed provider for reproducible Agent Skills sources (https://agentskills.io)
OCI registry provider for the Agent Skills format (https://agentskills.io)
Durable agent sessions for Dex, shared by the Dex agent framework adapters
Durable Claude Agent SDK sessions on Dex
Run Google ADK agents as durable Dex agent sessions
Durable Dex agent sessions for LangChain create_agent agents and LangGraph graphs
Run OpenAI Agents SDK agents as durable Dex agent sessions
Run Pydantic AI agents as durable Dex agent sessions
A Django Ninja-style abstraction for binding Django models to Pydantic AI Agents
Connector conformance suite + reference mock connector for the DocSpera Appointment Agent
Versioned data contracts (pydantic models + JSON Schema) for the DocSpera Appointment Agent
2026 Production Architecture Blueprint for FastAPI + Google Gemini Structured JSON (Pydantic v2, 3-Model Failover & SaaS Credit Quotas)
Model-agnostic LangChain agent middleware that retries with a corrective nudge when the model returns no text and no tool calls.
Model-agnostic LangChain agent middleware that detects unfilled template placeholders and forbidden internal terms in output and tool arguments.
Typed Valen decisions and uncertainty-aware routing for LangChain and LangGraph
Open Entity tools for CrewAI: adopt an identity, read it, record stances and arcs, verify and trust other agents.
Open Entity tools for the OpenAI Agents SDK: adopt an identity, read it, record stances and arcs, verify and trust other agents.
Minimal and fast AI agents for Python, with a Rust core
rboto: async-only, typed Python client for Amazon Bedrock Agents, powered by Rust. Alpha preview.
rboto: async-only, typed Python client for Amazon Bedrock Agent Runtime, powered by Rust. Alpha preview.
SDK Python para conectar aplicaciones al Canal API de agentes SayToken
Durable Claude Agent SDK runs on Temporal
AI-team marketplace and Build Library for Grok Bot, Claude Code, ChatGPT, Ollama, LM Studio, Open WebUI, n8n, CrewAI, LangGraph & MCP. Find teams, run them, fork them, or sell yours — with evidence-first runtime and Verified vs Untested status.
AI agent platform with BYOK/API support for indie builders and developers. Open ache.lol for current product details.
Free open-source desktop alternative to Grok Bot: local multi-agent Electron app powered by the Cursor SDK. Product page: https://tecadrise.ai/products/tecadrisebot Repo: https://github.com/tecadrise-ai/TecAdRiseBot
Hosted decision layer for AI agents (Choice, Score, Noul) for tool routing, ranking, and gated checks. Early-stage OSS TS SDK; ~2 stars.
Hosted decision layer for AI agents (Choice, Score, Noul) for tool routing, ranking, and gated checks. Early OSS TypeScript SDK @system-one-ai/core; ~2 GitHub stars.
Jev API gives software fast, typed decisions: send the app state, define the possible outcomes, and get back probabilities your code can act on. There is a free model (jev-1.13-free) to start with; API access and accounts are handled through BeatAPI.
Discover CharaxAI, an innovative AI agent platform.
Support for Agent Plugins: vendor-neutral manifest types and validators, installation, activation, digests and trusted contribution loading. Every Jini agent plugin ships here as its own subpath (@jini-ai/agent-plugins/<id>); resolving one never loads ano
AI agent bridge for Decentraland ??give your agents a body in the metaverse
Alias for @open-mercato/cezar — so `npx cezar-cli` just works. Local cockpit for running and tracking AI agent tasks in your repo.
Code mode for Po sessions: discover and exec over namespaces other packages contribute
LangChain integration for Agent-DID identities with middleware and tools for LangChain JS 1.x agents.
TypeScript runtime for in-app AI agents, with typed tools, configurable providers, approvals, and durable conversations.
LangChain tools for the Last Price pricing API.
Low-level primitives (loop, conversation, items, tools, streaming codecs) for building AI agents with Effect.
AgentVault OpenAI Codex bridge for encrypted owner messages and rooms
OpenAI engine for the Form.io Universal Agent Gateway (UAG): structured extraction, with research over read-only tools, on OpenAI or any server with its chat completions API (Ollama, vLLM, LM Studio, OpenRouter); and decisions with probabilities on OpenAI
JavaScript AI SDK for OpenAI, Claude, Gemini and Vertex AI: chatbots, AI agents, RAG with vector databases, tool calling, MCP, image and speech.
Shared types for Brightsy agents, chat, messages, tools, and plans
The lightweight foundation for building agents in TypeScript
OpenAI model provider for Tailored AI agents. Ships as a register(ctx) plugin — install with `tai plugin install @tailored-ai/provider-openai`; supersedes the minimal built-in `openai` provider with reasoning-model handling, org/project headers, and extra
Zero-dependency AI chaining and autonomous agent framework utilizing device resources for Node.js ESM and Android Termux
Framework-neutral generative UI for assistant-ui
Fayz SDK — plugin-inventory plugin
Official TypeScript SDK for Maritime: provision and drive AI agents on Maritime's serverless infrastructure from your own backend.
Write extensions for Ostia: the client library, its types, manifest schemas and a project generator
[experimental] Fayz SDK — plugin-dashboard plugin
Fayz SDK — API client, app params, runtime broker helpers, and shared types for generated Fayz projects
Capability-based assistant runtime with per-surface chat, transcripts, settings, and action tools.
Eve runtime adapter for assistant-ui
Fayz SDK — plugin-forms plugin
Zero-dependency, provider-agnostic agent harness: tool loop, guardrails, multi-agent orchestration, workflow graphs and sessions. No network or filesystem code in the core.
hmharness Cognitive OS core: Environment protocol, World Model, Goal system, Exploration, RLM runtime, five-layer memory, Skill Compiler, Continual Learning control plane, Evolution 2.0 governance, Multi-Agent topology, Transfer lab and GeneralBench schem
Efferent composable agent framework, 0.22.0. Node 24 or Bun.
Efferent composable agent framework, 0.22.0. Node 24 or Bun.
The fast AI agent framework for TypeScript and Node.js, a LangChain and LangGraph alternative on a Rust engine: tool calling, multi-agent workflows, human approval, durable threads, memory and an LLM gateway
Official CLI for the AGBoss TypeScript AI agent framework — project scaffolding (agboss create), code generators, static graph & saga verification (agboss check), describe --json, chat, and eval.
Headless client core for the LogicPair Agent: framework-free domain models behind createAgentClient, a React binding, and the same-page Host Bridge. Build a Custom UI on LogicPair's versioned client contract.
agentic-pi control plane: the single control.db writer (store), JSON-lines RPC server, peer auth, dispatcher, scheduler, the pi-control daemon entrypoint, and the ControlClient library (local + unix-socket).
Reproducible package manager for agentic AI skills (native binary launcher)
agentic-pi operator CLI (`org`): the human control-host tool for the approval inbox and other operator verbs, speaking to the control plane over its Unix socket with an admin token.
Agent frameworks in Restate virtual objects, their state in the object's SQL database: pi, OpenCode, LangGraph, Claude (Anthropic SDK) and DeepSeek Harness; and agentlib, journaled
Agent Instruction Protocol: the procedure language and its compiler, the runtime, the server and the client, as one `aip` command
Python client for agent-runtime over the standard library: generate_object() with answers validated against a full JSON Schema, generate_text(), and the admin API.
Chaos engineering for tool-using AI agents: deterministic fault injection, hash-chained traces, invariant verdicts, and policy replay.
Agent Compiler — Contract is the single source of truth. Arabic-first, framework-agnostic.
Forge Agent Bridge — universal agent session/event/approval contracts and provider adapters (Claude Code, Codex, Devin)
HarnessWeaver plugin: Anthropic models.
The file tools a platform agent gets (read / search / create / render / edit an artifact through file-artifact), the extended knowledge-base toolset (FR-180), and how a turn's attachments reach the model. One definition for visual-builder and master-agent, so the two cannot drift.
An integration package connecting Cleave and LangChain
Thin Protocols for LLM work (Generator, Embedder, Reranker, Tracer, PromptSource, Judge, ...) and the one call path around them: roles from a tier table, prompt packs with a frozen content sha, per-call graders, named outcomes. Frameworks live behind the ports; pydantic-ai is the first adapter.
Remote Project Agent command-line client
Pydantic models and a client for the ai-agent-subsystem CRDs (Agent / Station / AgentDefinition / ModelCatalog / AgentSpawner). Types are generated from the D source so they cannot drift.
Run an agent on StackShift: events, governed actions, StackShift calls, state and model access.
The Pythonic agentic framework for TensorZero
Alias of apex-fusion-agent-sdk, the Python SDK for Vector (import package vector_agent).
Genspark is an all-in-one AI workspace rather than a single agent: its product line spans Genspark Claw, Speakly, GenClipboard and GenTerminal, and it operates regional editions across several countries. The directory also holds its published npm CLI separately; this entry is the product. Catalogued from the vendor's own site; MeshKore does not operate it.
Mega Desk handoff/gates kit for fleets already spending on seats. Free handoff: https://www.agensi.io/skills/agent-handoff-lite Three-gate ops ($12): https://www.agensi.io/skills/three-gate-ops Evidence-bus ($9): https://www.agensi.io/skills/evidence-bus-starter