Agents

Agents in almyty are composable AI pipelines that orchestrate LLM calls, tool executions, conditional logic, and data transformations into repeatable workflows. Each agent is defined as a directed acyclic graph (DAG) of nodes connected by edges.

Key Concepts

Pipelines

A pipeline is the core structure of an agent. It consists of:

Nodes — individual processing steps (LLM calls, tool invocations, transforms, etc.)
Edges — connections that define data flow between nodes

Every pipeline starts with an Input node and ends with an Output node. Between them, you wire up any combination of the 9 available node types.

Execution Model

When an agent is invoked, almyty:

Validates the input against the Input node's schema
Topologically sorts the pipeline nodes
Executes each node in order, passing data along edges
Resolves template expressions ({{nodes.llm_1.output}}) at runtime
Returns the Output node's result

Execution is synchronous by default. Each node runs only after all its upstream dependencies have completed.

Agent Lifecycle

Status	Description
`draft`	Agent is being built, not yet invokable
`active`	Agent is live and accepting invocations
`inactive`	Agent is paused, invocations will be rejected
`error`	Agent has a configuration issue

Creating an Agent

Via the UI

Navigate to Agents and click Create Agent
Give it a name and optional description
The default pipeline has Input -> LLM Call -> Output nodes
Use the visual editor to add, remove, and connect nodes
Configure each node by clicking on it in the canvas

Via the API

curl -X POST https://api.almyty.com/agents \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Research Agent",
    "description": "Searches and summarizes information",
    "pipeline": {
      "nodes": [
        { "id": "input_1", "type": "input", "data": {
          "schema": { "type": "object", "properties": { "query": { "type": "string" } }, "required": ["query"] }
        }},
        { "id": "llm_1", "type": "llm_call", "data": {
          "providerId": "provider-uuid",
          "systemPrompt": "You are a research assistant.",
          "userPromptTemplate": "Research: {{input.query}}"
        }},
        { "id": "output_1", "type": "output", "data": {
          "mapping": "{{nodes.llm_1.output}}"
        }}
      ],
      "edges": [
        { "id": "e1", "source": "input_1", "target": "llm_1" },
        { "id": "e2", "source": "llm_1", "target": "output_1" }
      ]
    }
  }'

From a Template

almyty ships with several built-in templates:

Template	Description
Simple Chat	Input -> LLM -> Output. The simplest possible agent.
Research Agent	Chains multiple LLM calls with web search tools.
Tool-Augmented	LLM with tool calling for dynamic API interaction.
Multi-Step Pipeline	Conditional branching with data transformations.

Select a template when creating an agent to get a pre-wired pipeline.

Invoking an Agent

curl -X POST https://api.almyty.com/agents/{id}/invoke \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "input": { "query": "What is the weather in Berlin?" }
  }'

Response:

{
  "success": true,
  "data": {
    "output": "The current weather in Berlin is 12°C with partly cloudy skies.",
    "executionId": "exec-uuid",
    "duration": 2340,
    "tokensUsed": 156
  }
}

Import & Export

Export an agent as JSON for version control or sharing:

# Export
curl https://api.almyty.com/agents/{id}/export \
  -H "Authorization: Bearer $TOKEN" > agent.json
 
# Import
curl -X POST https://api.almyty.com/agents/import \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d @agent.json

Tool Scoping Node Types