MAQPNA

Compare

Agent frameworks vs an agent runtime.

Frameworks like LangGraph, LangChain, CrewAI, the OpenAI Agents SDK and the Claude Agent SDK are how you build an agent.

Different layers, different jobs

Agent frameworkAgent runtime (MAQPNA)
Main jobBuild the agent: prompts, planning, tools, memory, orchestrationRun the agent: isolation, identity, policy, approval, audit
Where it runsInside the agent's processAround the agent: a sandbox per session and a gateway in front of every tool
GuardrailsChecks in the same process as the model's decisions, so they are only as strong as that processEnforced at the gateway, outside the agent; the agent cannot skip them
Human approvalMany frameworks can pause for human input within the applicationA policy rule holds the call at the gateway; separation of duties and four-eyes; recorded in the ledger
CredentialsUsually the application's API keys, available to the agentA short-lived identity per session; upstream keys injected at the gateway, never seen by the agent
EvidenceApplication logs and tracingA hash-chained, verifiable audit ledger, plus OpenTelemetry traces
IsolationNot in scopegVisor, microVM or confidential VM per session, default-deny networking

Use them together

Keep your framework.

terminal
pip install "kosa8-maqpna[langgraph]"
maqpna dev up
maqpna dev run -- python my_langgraph_agent.py
maqpna dev timeline --last

FAQ

Common questions

Is MAQPNA an agent framework?

No.

Do I have to rewrite my agent?

No.

Can I keep my framework's own guardrails?

Yes.

Keep your framework. Add a runtime.

Run your existing agent under a local MAQPNA with one command and see every call it makes.

More install options
curl -fsSL https://maqpna.com/install.sh | sh