> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mithunai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# MithunAI Agent Studio

> Build stateful multi-agent workflows with tool authorization, directed graph execution, and human-in-the-loop approval gates.

Agent Studio executes deterministic multi-agent state graphs with tool calling, human-in-the-loop approvals, and query-time permission verification.

Tasks that exceed a single prompt-response turn are decomposed into graph nodes executed across specialist agents.

```
User Query ──> [ Orchestrator Agent ]
                     │
         ┌───────────┴───────────┐
         ▼                       ▼
  [ Research Agent ]     [ Synthesis Agent ]
   (search_knowledge)    (Generates Draft)
         │                       │
         └───────────┬───────────┘
                     ▼
           [ Verification Agent ]
          (Checks Citations / Invariants)
                     │
                     ▼
           Verified Final Answer
```

## Graph Primitives

<Columns cols={3}>
  <Card title="State Graphs" icon="diagram-project">
    Define execution DAGs with branching logic, error recovery pathways, and typed state
    transitions.
  </Card>

  <Card title="Tool Calling" icon="wrench">
    Equip agents with verified read-only MCP tools or custom API integrations that require signed
    session authorization.
  </Card>

  <Card title="Human Approval Gates" icon="user-check">
    Pause stateful workflows before executing sensitive actions until an authorized operator
    approves the turn.
  </Card>
</Columns>

## Agent Security Boundaries

1. **Non-Escalating Scopes**: Agents execute under the permissions of the calling principal. An agent cannot call endpoints or access collections forbidden to the initiating user or service key.
2. **Deterministic Step Limits**: Workflows enforce hard recursion and turn budgets (default: 15 turns) to prevent infinite reasoning loops.
3. **Audited Tool Payloads**: Every parameter passed to external tools and every response returned is persisted to the immutable audit log.