---
title: "ProcessorRegistry"
manual: "AI Assistant"
version: "main"
source: "Developers/Registries/ProcessorRegistry.rst"
rendered: "2026-10-03T14:54:45+00:00"
---

# ProcessorRegistry {#processorregistry}

## Purpose {#purpose}

Resolves assistant pipeline processors by identifier.

## Required interface {#required-interface}

`Madj2k\AiCore\Assistant\Pipeline\Processor\ProcessorInterface`

## Service registration {#service-registration}

Register implementations through a service tag:

```yaml
services:
  Vendor\Extension\Assistant\Pipeline\Processor\MyProcessor:
    tags:
      - name: 'aiassistant.assistant.pipeline.processor'
```

## Identifier {#identifier}

The implementation should expose a stable identifier. Configuration records use
that identifier to resolve the implementation at runtime.

## Implementation rules {#implementation-rules}

-   Keep services stateless where possible.
-   Read credentials and runtime options from configuration records.
-   Throw meaningful exceptions for invalid configuration.
-   Return core DTOs rather than provider-specific arrays where interfaces
    define DTOs.

## Processor type matching {#processor-type-matching}

The registry resolves processors by `processor_identifier`. The pipeline step
`type` is still important and should always be one of the values defined by
`Madj2k\AiCore\Assistant\Enum\AssistantPipelineProcessorType`.

Processor implementations should use this enum when implementing `supports()`
or equivalent matching logic. Static SQL and default configurations should also
use enum values for the `type` column.