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

# Agent

> Define an AI agent with a model, tools, knowledge, and memory

> Configures an Agno AI agent with all its dependencies — model, tools, knowledge base, memory, and storage.

## Config

| Field                      | Type               | Required | Default       | Description                                                  |
| -------------------------- | ------------------ | -------- | ------------- | ------------------------------------------------------------ |
| `name`                     | `string`           | No       | Resource name | Display name for the agent                                   |
| `description`              | `string`           | No       | —             | Agent description                                            |
| `role`                     | `string`           | No       | —             | Agent role description                                       |
| `model`                    | `Dependency`       | **Yes**  | —             | Reference to `agno/models/anthropic` or `agno/models/openai` |
| `instructions`             | `list[string]`     | No       | —             | System instructions for the agent                            |
| `prompt`                   | `Dependency`       | No       | —             | Reference to `agno/prompt` for reusable instructions         |
| `tools`                    | `list[Dependency]` | No       | `[]`          | References to `agno/tools/mcp` or `agno/tools/websearch`     |
| `knowledge`                | `Dependency`       | No       | —             | Reference to `agno/knowledge`                                |
| `db`                       | `Dependency`       | No       | —             | Reference to `agno/db/postgres` for session storage          |
| `memory`                   | `Dependency`       | No       | —             | Reference to `agno/memory/manager`                           |
| `markdown`                 | `boolean`          | No       | `false`       | Format responses as Markdown                                 |
| `add_datetime_to_context`  | `boolean`          | No       | `false`       | Include current datetime in context                          |
| `read_chat_history`        | `boolean`          | No       | Auto          | Read chat history from DB (auto-enabled when `db` is set)    |
| `add_history_to_context`   | `boolean`          | No       | Auto          | Add history to context (auto-enabled when `db` is set)       |
| `num_history_runs`         | `integer`          | No       | —             | Number of previous runs to include in history                |
| `enable_agentic_memory`    | `boolean`          | No       | `false`       | Enable agentic memory capabilities                           |
| `update_memory_on_run`     | `boolean`          | No       | `false`       | Update memory after each run                                 |
| `add_memories_to_context`  | `boolean`          | No       | —             | Add stored memories to context                               |
| `enable_session_summaries` | `boolean`          | No       | `false`       | Enable session summary generation                            |

## Outputs

| Field              | Type           | Description                                               |
| ------------------ | -------------- | --------------------------------------------------------- |
| `spec`             | `object`       | Serialized agent specification for runtime reconstruction |
| `pip_dependencies` | `list[string]` | Python packages required by agent tools and knowledge     |

## Dependencies

**Depends on:**

* `agno/models/anthropic` or `agno/models/openai` (required) — the LLM powering the agent
* `agno/tools/mcp` or `agno/tools/websearch` (optional) — tool integrations
* `agno/knowledge` (optional) — semantic search knowledge base
* `agno/memory/manager` (optional) — persistent memory
* `agno/db/postgres` (optional) — session and chat history storage
* `agno/prompt` (optional) — reusable prompt template

**Depended on by:**

* `agno/team` — as a team member
* `agno/runner` — for deployment to Kubernetes

## Example

```yaml theme={"theme":{"light":"min-light","dark":"min-dark"}}
provider: agno
resource: agent
name: my-assistant
config:
  description: "A helpful AI assistant"
  model:
    provider: agno
    resource: models/anthropic
    name: claude
  instructions:
    - "You are a helpful assistant."
    - "Always respond concisely."
  tools:
    - provider: agno
      resource: tools/websearch
      name: search
  markdown: true
```

## Notes

* The agent resource is **stateless** — it produces a serializable spec used by `agno/runner` for deployment.
* When `db` is set, `read_chat_history` and `add_history_to_context` are automatically enabled unless explicitly overridden.
* Updating a dependency (e.g., changing the model) triggers the agent to rebuild automatically.