# Connector reference

> Source: https://docs.clonepartner.com/reference/connectors/

Connectors adapt external databases, files, object stores, and application APIs to Envoy tasks. The deployed build determines the available connector types and exposes each type's configuration schema.

## Configuration ownership

A task declares only a logical slot:

```yaml
connectors:
  source:
    type: mongodb
```

A stored connector or CLI connector configuration supplies endpoint, path, account, and credentials. A job template maps its connector names to task slots, and a job binds those names to stored connector records.

## Connector categories

### Databases

- `sqlite`
- `postgres`
- `mysql`
- `mssql`
- `mongodb`

Database connectors provide methods appropriate to their model, such as queries, table or collection reads, lookups, counts, inserts, updates, and upserts.

### Files

- `csv`
- `yaml`

File connectors read and, where supported, write structured files. Paths must exist inside the executor's filesystem namespace.

### Object storage

- `s3`
- `azure-blob`

Object-store connectors list, read, write, and transfer objects according to configured permissions.

### Application APIs

Available built-in or selected application connectors can include:

- `truto`
- `dynamics365`
- `jira`
- `freshdesk`

Customer builds may register additional catalog-powered connector types.

## Common behavior

Method names and arguments vary by type. Use the connector type schema and introspection API as the source for the deployed build.

Collection methods should:

- stream pages in task pipelines;
- return one page plus an opaque cursor for UI discovery;
- support server-side search where the upstream API permits it;
- avoid buffering the full collection.

## Discovery

Connector introspection can expose resources, fields, connected accounts, or method-specific option lists. Discovery methods must be read-only and bounded.

Large picker lists use server search and `next_cursor`. Do not fetch all options and filter them only in the browser.

## File uploads

Connector types that support managed uploads expose that capability in the connector editor. The stored path or key is deployment-specific. Ensure both control plane and executor can access it.

## Logging and errors

Connector logs are associated with the run. Avoid logging headers, tokens, connection strings, or full sensitive records. Preserve provider request IDs and status codes when they help support a failure.

## Availability and portability

Task bundles include connector slot names and type requirements but no connector records. Before import:

1. verify every type exists in the target build;
2. create and test target connectors;
3. bind connector names during job creation;
4. run a bounded read and write test.

See [Catalog connectors and connected accounts](/connectors/catalog-and-connected-accounts) and [Truto connector](/connectors/truto).