> For the complete documentation index, see [llms.txt](https://docs.fastn.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.fastn.ai/ai-agents/configuration-agent.md).

# Configuration Agent

The Configuration Agent is a standalone agent that handles the most tedious part of integration setup — configuring how data flows between systems. Instead of manually inspecting API responses, writing filter conditions, and testing configurations, the agent probes live APIs and guides you through the setup interactively.

### What it does

1. **Probes live APIs** — Connects to the third-party app using the tenant's credentials and fetches sample data. This gives you real examples of what the data looks like — field names, data types, value formats, and edge cases.
2. **Presents sample data** — Shows you actual records from the connected app so you can make informed decisions about filtering and mapping.
3. **Helps define filter conditions** — Based on the sample data, helps you set up filters ("only sync orders where status is 'completed'", "only sync contacts with email addresses").
4. **Configures targets** — Defines where data should go and how it should be transformed — which fields map to the target system, what format conversions are needed, and what defaults to apply for missing values.
5. **Handles edge cases** — Detects configuration scenarios that could cause problems: required fields that are sometimes empty, date formats that vary between records, currency precision mismatches, enum values that don't map cleanly.

### When it's used

The Configuration Agent is called when:

* Setting up field mappings between two connectors for the first time
* A flow needs conditional logic based on real data patterns
* A tenant has complex data with edge cases (mixed date formats, optional fields, multi-currency)
* You need to debug why synced data doesn't look right in the target system

### How it works in practice

**You:** "I need to configure the Shopify-to-Xero order sync. Only completed orders, and map the line items correctly."

**Agent:** "Let me pull some sample orders from the connected Shopify account..."

*\[Agent probes the Shopify API and returns sample data]*

**Agent:** "Here are 5 recent orders. I can see that:

* Orders have 4 possible statuses: `pending`, `processing`, `completed`, `cancelled`
* Line items include `variant_id`, `title`, `quantity`, `price` — price is a string, not a number
* Shipping addresses sometimes have an empty `company` field
* Currency is CAD for this account, not USD

Based on this, I'd recommend:

* Filter: `status == 'completed'`
* Transform: convert `price` from string to number before mapping to Xero
* Default: use the customer name when `company` is empty
* Currency: map `CAD` to Xero's currency code

Want me to apply this configuration?"

> **📷 Screenshot needed:** Configuration Agent showing sample data from a live API with its analysis and recommended configuration.

> **📷 Screenshot needed:** The agent presenting filter conditions it has generated based on the sample data.

### What it detects

The agent actively looks for configuration pitfalls:

| Issue                       | What the agent does                                                                                    |
| --------------------------- | ------------------------------------------------------------------------------------------------------ |
| **Type mismatches**         | Price as string vs number, dates as Unix timestamp vs ISO format — flags and offers transform          |
| **Missing required fields** | Fields that are required in the target but sometimes empty in the source — suggests defaults           |
| **Enum mapping gaps**       | Status values in the source that don't have equivalents in the target — asks you to define the mapping |
| **Currency precision**      | Source uses 2 decimal places, target uses 4 — suggests rounding strategy                               |
| **Locale differences**      | Date formats (MM/DD vs DD/MM), number separators (comma vs period) — configures the Type Coercer       |
| **Array/nested data**       | Line items, addresses, tags that need flattening or restructuring — sets up the Field Mapper           |

### Configuration output

The Configuration Agent produces configurations that slot into Fastn's 5-level hierarchy:

* **Field mappings** — source field → CDM field → target field, with transforms
* **Filter conditions** — condition AST compatible with the Flow Router (operators: eq, neq, gt, lt, in, contains, exists)
* **Type coercion rules** — date format handling, number locale, currency precision, boolean mapping
* **Default values** — fallbacks for optional fields

These configurations are saved at the appropriate level — typically **integration** or **tenant** scope, but can be overridden at the flow level for specific cases.

### Related pages

* [**Configuring a Connector**](https://claude.ai/tutorials/saas-admin/configuring-a-connector.md) — Manual connector configuration tutorial
* [**Connector Agent**](https://claude.ai/chat/connector-agent.md) — Builds the connectors that the Configuration Agent configures
* [**Tenant Onboarding Agent**](https://claude.ai/chat/tenant-onboarding.md) — Uses the Configuration Agent's capabilities when helping tenants set up field mappings


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.fastn.ai/ai-agents/configuration-agent.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.