On‑chain autonomous agents are software agents that combine large language models, stateful orchestration, and blockchain primitives to act, transact, and persist decisions on-chain. This topic covers the developer toolkits and marketplaces that make those agents practical: agent frameworks for orchestration and state management, document/RAG layers for grounding, no‑code/enterprise platforms for governance, and decentralized infrastructure for secure execution and monetization via marketplaces such as OKX OnchainOS. As of 2026, real deployments and marketplace activity have increased demand for composable, auditable toolchains. LangChain remains central for building, testing, and deploying stateful agent pipelines (including LangGraph-style state management and debugging). LlamaIndex provides production-ready document agents and retrieval-augmented generation (RAG) primitives to ground agent decisions in indexed on‑chain or off‑chain data. AutoGPT-style platforms offer patterns for running autonomous workflows—self‑hosted or cloud—useful for continuous on‑chain automation. StackAI and similar no‑code/low‑code platforms address enterprise needs for deployment, access controls, and governance. Specialized AI‑integrated wallet development patterns focus on secure key usage, transaction automation, and UX integration for agent-driven wallets. Key concerns driving toolkit selection are security and key management, reproducibility and evaluation, gas and compute efficiency, privacy and data locality, and marketplace economics (tokenized access, billing). Effective stacks combine an orchestration layer (LangChain/AutoGPT patterns), a retrieval/document layer (LlamaIndex), governance/no‑code tooling (StackAI), and secure wallet/infra integration for on‑chain execution. This blended approach helps developers move from prototypes to auditable, marketplace‑ready agents on platforms like OKX OnchainOS.