AI Agents
Pre-built agents you can install, inspect, and run. Every agent declares what it does, what it needs, and how it behaves — in a single, standardized manifest.
Available Agents
What is an AgentNode Agent?
An AgentNode agent is a self-describing AI agent packaged with a standardized manifest. Unlike traditional AI agents that hide their behavior in code, AgentNode agents declare everything upfront: their goal, behavior, tool access, permissions, and limits.
This means you can inspect before you install. You know exactly what the agent does, which tools it uses, what API keys it needs, and what permissions it requires — before running a single line of code.
Agents are composable. They use AgentNode tool packs as building blocks — a research agent might combine web search, page extraction, and document summarization, all installed from the registry.
Each agent is automatically verified by AgentNode before listing: installed, imported, and smoke-tested in a sandbox. You see the verification score, tier, and any issues upfront.
Agent Tiers
Agents are classified by what they need to run. This tells you at a glance how much trust and configuration is required.
LLM Only
Pure reasoning agents. They use your LLM to think, write, and plan — no external tools or API calls needed.
Examples: Blog writer, newsletter, project planner, report generator
LLM + Tools
Agents that combine LLM reasoning with AgentNode tool packs. They search the web, extract documents, analyze data, and more.
Examples: Deep research, code review, competitive intel, fact checker
LLM + Credentials
Agents that connect to external services using API keys or OAuth. They interact with your CRM, cloud provider, email, or databases.
Examples: CRM enrichment, cloud cost analysis, email triage, deployment
How It Works
Install
One command. The agent and all its dependencies are downloaded, verified, and installed locally.
agentnode install deep-research-agent
Configure
Each agent declares what it needs: API keys, permissions, tool access. You see everything upfront — no hidden behavior.
result = run_tool("deep-research-agent",
goal="Compare React vs Vue in 2026")Run
Agents run locally on your machine with declared isolation. They orchestrate tool calls, handle errors, and return structured results.
print(result["report"]) print(result["sources"])
Why AgentNode Agents?
Standardized Manifest
Every agent declares its goal, behavior, permissions, tool access, and limits in a single agentnode.yaml. No guessing what an agent does or needs.
Verified Before You Install
Each agent is automatically installed, imported, and smoke-tested by AgentNode before listing. You see the verification score and tier upfront.
Transparent Behavior
The agent's behavior description, tool access, system prompt, and permission levels are all visible on the package page. Nothing is hidden.
Any LLM Provider
Agents work with any LLM provider — OpenAI, Anthropic, Gemini, or OpenRouter. The agent uses the same model that invokes it, auto-detected from your API key.
Declared Permissions
Network access, filesystem access, code execution — every permission level is declared in the manifest and shown before installation.
Composable
Agents use AgentNode tool packs as building blocks. A research agent combines web search, page extraction, and summarization — all from the registry.
Build Your Own Agent
Create an agent with a standardized manifest, publish it to the registry, and let others install and run it. The manifest format is open and documented.