> For the complete documentation index, see [llms.txt](https://docs.curiostorage.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.curiostorage.org/market-2.0/products/pdp_v1.md).

# PDP v1

## What `pdp_v1` Is For

`pdp_v1` drives PDP dataset and piece lifecycle operations through MK20.

## Struct Definition

```go
type PDPV1 struct {
    CreateDataSet bool    `json:"create_data_set"`
    DeleteDataSet bool    `json:"delete_data_set"`
    AddPiece      bool    `json:"add_piece"`
    DeletePiece   bool    `json:"delete_piece"`
    DataSetID     *uint64 `json:"data_set_id,omitempty"`
    RecordKeeper  string  `json:"record_keeper"`
    PieceIDs      []uint64 `json:"piece_ids,omitempty"`
    ExtraData     []byte   `json:"extra_data,omitempty"`
}
```

## Action Model

Exactly one of these flags must be set in each deal:

* `create_data_set`
* `delete_data_set`
* `add_piece`
* `delete_piece`

## Key Fields

* `data_set_id`: required for all actions except `create_data_set`.
* `record_keeper`: required when creating a dataset.
* `piece_ids`: required for `delete_piece`.
* `extra_data`: verifier/service payload. It is required for `create_data_set`, `add_piece`, and `delete_piece`, and must be absent for `delete_data_set`.

## Validation Rules

* `pdp_v1` must be enabled by provider.
* Provider must have PDP signing key configured (`eth_keys.role = 'pdp'`).
* Only one action flag is allowed per deal.
* `create_data_set`:
  * `data_set_id` must be absent
  * `record_keeper` must be a valid hex address
  * `extra_data` must be present and no larger than 4 KiB
* `delete_data_set` / `add_piece`:
  * `data_set_id` must be present
  * dataset must exist and be active
* `delete_data_set`:
  * `extra_data` must be absent
* `add_piece`:
  * `extra_data` must be present and no larger than 8 KiB
* `delete_piece`:
  * `data_set_id` must be present
  * `piece_ids` must be present
  * all listed pieces must exist and be active in that dataset
  * `extra_data` must be present and no larger than 256 bytes

## Runtime Sanitize Rules

* `add_piece` requires `retrieval_v1`.
* `offline` source is rejected for PDP flows.
* Raw-format data cannot use `announce_payload`.
* Dataset and piece ownership checks are revalidated.

## Processing Behavior

* `add_piece` with HTTP/aggregate data goes directly into PDP processing pipelines.
* `add_piece` with `put` source (or upload-first mode) enters upload-waiting first.
* Create/delete operations enqueue PDP lifecycle actions without upload flow.