Restrict the changes to the new preprocessing API only.

docs/_static/llama-stack-spec.html

@@ -8052,44 +8052,6 @@
                 ],
                 "title": "HealthInfo"
             },
-            "PreprocessorChainElement": {
-                "type": "object",
-                "properties": {
-                    "preprocessor_id": {
-                        "type": "string"
-                    },
-                    "options": {
-                        "type": "object",
-                        "additionalProperties": {
-                            "oneOf": [
-                                {
-                                    "type": "null"
-                                },
-                                {
-                                    "type": "boolean"
-                                },
-                                {
-                                    "type": "number"
-                                },
-                                {
-                                    "type": "string"
-                                },
-                                {
-                                    "type": "array"
-                                },
-                                {
-                                    "type": "object"
-                                }
-                            ]
-                        }
-                    }
-                },
-                "additionalProperties": false,
-                "required": [
-                    "preprocessor_id"
-                ],
-                "title": "PreprocessorChainElement"
-            },
             "RAGDocument": {
                 "type": "object",
                 "properties": {
@@ -8171,12 +8133,6 @@
                     },
                     "chunk_size_in_tokens": {
                         "type": "integer"
-                    },
-                    "preprocessor_chain": {
-                        "type": "array",
-                        "items": {
-                            "$ref": "#/components/schemas/PreprocessorChainElement"
-                        }
                     }
                 },
                 "additionalProperties": false,
@@ -9382,6 +9338,44 @@
                 ],
                 "title": "PreprocessingDataElement"
             },
+            "PreprocessorChainElement": {
+                "type": "object",
+                "properties": {
+                    "preprocessor_id": {
+                        "type": "string"
+                    },
+                    "options": {
+                        "type": "object",
+                        "additionalProperties": {
+                            "oneOf": [
+                                {
+                                    "type": "null"
+                                },
+                                {
+                                    "type": "boolean"
+                                },
+                                {
+                                    "type": "number"
+                                },
+                                {
+                                    "type": "string"
+                                },
+                                {
+                                    "type": "array"
+                                },
+                                {
+                                    "type": "object"
+                                }
+                            ]
+                        }
+                    }
+                },
+                "additionalProperties": false,
+                "required": [
+                    "preprocessor_id"
+                ],
+                "title": "PreprocessorChainElement"
+            },
             "PreprocessRequest": {
                 "type": "object",
                 "properties": {

docs/_static/llama-stack-spec.yaml

@@ -5572,25 +5572,6 @@ components:
       required:
         - status
       title: HealthInfo
-    PreprocessorChainElement:
-      type: object
-      properties:
-        preprocessor_id:
-          type: string
-        options:
-          type: object
-          additionalProperties:
-            oneOf:
-              - type: 'null'
-              - type: boolean
-              - type: number
-              - type: string
-              - type: array
-              - type: object
-      additionalProperties: false
-      required:
-        - preprocessor_id
-      title: PreprocessorChainElement
     RAGDocument:
       type: object
       properties:
@@ -5639,10 +5620,6 @@ components:
           type: string
         chunk_size_in_tokens:
           type: integer
-        preprocessor_chain:
-          type: array
-          items:
-            $ref: '#/components/schemas/PreprocessorChainElement'
       additionalProperties: false
       required:
         - documents
@@ -6414,6 +6391,25 @@ components:
       required:
         - data_element_id
       title: PreprocessingDataElement
+    PreprocessorChainElement:
+      type: object
+      properties:
+        preprocessor_id:
+          type: string
+        options:
+          type: object
+          additionalProperties:
+            oneOf:
+              - type: 'null'
+              - type: boolean
+              - type: number
+              - type: string
+              - type: array
+              - type: object
+      additionalProperties: false
+      required:
+        - preprocessor_id
+      title: PreprocessorChainElement
     PreprocessRequest:
       type: object
       properties:

docs/source/distributions/remote_hosted_distro/nvidia.md

@@ -10,7 +10,6 @@ The `llamastack/distribution-nvidia` distribution consists of the following prov
 | eval | `inline::meta-reference` |
 | inference | `remote::nvidia` |
 | post_training | `remote::nvidia` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `remote::nvidia` |
 | scoring | `inline::basic` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/bedrock.md

@@ -16,7 +16,6 @@ The `llamastack/distribution-bedrock` distribution consists of the following pro
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::bedrock` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `remote::bedrock` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/cerebras.md

@@ -9,7 +9,6 @@ The `llamastack/distribution-cerebras` distribution consists of the following pr
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::cerebras`, `inline::sentence-transformers` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/fireworks.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-fireworks` distribution consists of the following p
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::fireworks`, `inline::sentence-transformers` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/groq.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-groq` distribution consists of the following provid
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::groq` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/meta-reference-gpu.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-meta-reference-gpu` distribution consists of the fo
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `inline::meta-reference` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/meta-reference-quantized-gpu.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-meta-reference-quantized-gpu` distribution consists
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `inline::meta-reference-quantized` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/ollama.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-ollama` distribution consists of the following prov
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::ollama` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/remote-vllm.md

@@ -18,7 +18,6 @@ The `llamastack/distribution-remote-vllm` distribution consists of the following
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::vllm`, `inline::sentence-transformers` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/sambanova.md

@@ -17,7 +17,6 @@ The `llamastack/distribution-sambanova` distribution consists of the following p
 |-----|-------------|
 | agents | `inline::meta-reference` |
 | inference | `remote::sambanova` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | telemetry | `inline::meta-reference` |
 | tool_runtime | `remote::brave-search`, `remote::tavily-search`, `inline::code-interpreter`, `inline::rag-runtime` |

docs/source/distributions/self_hosted_distro/tgi.md

@@ -20,7 +20,6 @@ The `llamastack/distribution-tgi` distribution consists of the following provide
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::tgi`, `inline::sentence-transformers` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

docs/source/distributions/self_hosted_distro/together.md

@@ -19,7 +19,6 @@ The `llamastack/distribution-together` distribution consists of the following pr
 | datasetio | `remote::huggingface`, `inline::localfs` |
 | eval | `inline::meta-reference` |
 | inference | `remote::together`, `inline::sentence-transformers` |
-| preprocessing | `inline::basic`, `inline::simple_chunking` |
 | safety | `inline::llama-guard` |
 | scoring | `inline::basic`, `inline::llm-as-judge`, `inline::braintrust` |
 | telemetry | `inline::meta-reference` |

llama_stack/apis/tools/rag_tool.py

@@ -11,7 +11,6 @@ from pydantic import BaseModel, Field
 from typing_extensions import Annotated, Protocol, runtime_checkable
 
 from llama_stack.apis.common.content_types import URL, InterleavedContent
-from llama_stack.apis.preprocessing import PreprocessorChain
 from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
 
@@ -87,7 +86,6 @@ class RAGToolRuntime(Protocol):
         documents: List[RAGDocument],
         vector_db_id: str,
         chunk_size_in_tokens: int = 512,
-        preprocessor_chain: Optional[PreprocessorChain] = None,
     ) -> None:
         """Index documents so they can be used by the RAG system"""
         ...

llama_stack/distribution/routers/routers.py

@@ -677,13 +677,12 @@ class ToolRuntimeRouter(ToolRuntime):
             documents: List[RAGDocument],
             vector_db_id: str,
             chunk_size_in_tokens: int = 512,
-            preprocessor_chain: Optional[PreprocessorChain] = None,
         ) -> None:
             logger.debug(
                 f"ToolRuntimeRouter.RagToolImpl.insert: {vector_db_id}, {len(documents)} documents, chunk_size={chunk_size_in_tokens}"
             )
             return await self.routing_table.get_provider_impl("insert_into_memory").insert(
-                documents, vector_db_id, chunk_size_in_tokens, preprocessor_chain
+                documents, vector_db_id, chunk_size_in_tokens
             )
 
     def __init__(

llama_stack/providers/inline/preprocessing/basic/__init__.py

@@ -1,18 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-from .config import InlineBasicPreprocessorConfig
-
-
-async def get_provider_impl(
-    config: InlineBasicPreprocessorConfig,
-    _deps,
-):
-    from .basic import InclineBasicPreprocessorImpl
-
-    impl = InclineBasicPreprocessorImpl(config)
-    await impl.initialize()
-    return impl

llama_stack/providers/inline/preprocessing/basic/basic.py

@@ -1,151 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-import logging
-import re
-from typing import Any, List, Optional
-
-import httpx
-
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.preprocessing import (
-    Preprocessing,
-    PreprocessingDataElement,
-    PreprocessingDataFormat,
-    PreprocessingDataType,
-    Preprocessor,
-    PreprocessorChain,
-    PreprocessorOptions,
-    PreprocessorResponse,
-)
-from llama_stack.providers.datatypes import PreprocessorsProtocolPrivate
-from llama_stack.providers.inline.preprocessing.basic.config import InlineBasicPreprocessorConfig
-from llama_stack.providers.utils.inference.prompt_adapter import interleaved_content_as_str
-from llama_stack.providers.utils.memory.vector_store import content_from_data, parse_pdf
-
-log = logging.getLogger(__name__)
-
-
-class InclineBasicPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate):
-    # this preprocessor can either receive documents (text or binary) or document URIs
-    input_types = [
-        PreprocessingDataType.binary_document,
-        PreprocessingDataType.raw_text_document,
-        PreprocessingDataType.document_uri,
-    ]
-
-    # this preprocessor optionally retrieves the documents and converts them into plain text
-    output_types = [PreprocessingDataType.raw_text_document]
-
-    preprocessor_store = None
-
-    URL_VALIDATION_PATTERN = re.compile("^(https?://|file://|data:)")
-
-    def __init__(self, config: InlineBasicPreprocessorConfig) -> None:
-        self.config = config
-
-    async def initialize(self) -> None:
-        pass
-
-    async def shutdown(self) -> None:
-        pass
-
-    async def register_preprocessor(self, preprocessor: Preprocessor) -> None:
-        pass
-
-    async def unregister_preprocessor(self, preprocessor_id: str) -> None:
-        pass
-
-    async def do_preprocess(
-        self,
-        preprocessor_id: str,
-        preprocessor_inputs: List[PreprocessingDataElement],
-        options: Optional[PreprocessorOptions] = None,
-    ) -> PreprocessorResponse:
-        results = []
-
-        for inp in preprocessor_inputs:
-            input_type = self._resolve_input_type(inp)
-
-            if input_type == PreprocessingDataType.document_uri:
-                document = await self._fetch_document(inp)
-                if document is None:
-                    continue
-            elif input_type == PreprocessingDataType.binary_document:
-                document = inp.data_element_path_or_content
-                if inp.data_element_format is None:
-                    log.error(f"Binary document format is not provided for {inp.data_element_id}, skipping it")
-                    continue
-                if inp.data_element_format != PreprocessingDataFormat.pdf:
-                    log.error(
-                        f"Unsupported binary document type {inp.data_element_format} for {inp.data_element_id}, skipping it"
-                    )
-                    continue
-            elif input_type == PreprocessingDataType.raw_text_document:
-                document = interleaved_content_as_str(inp.data_element_path_or_content)  # type: ignore
-            else:
-                log.error(f"Unexpected preprocessor input type: {input_type}")
-                continue
-
-            if inp.data_element_format == PreprocessingDataFormat.pdf:
-                document = parse_pdf(document)
-
-            new_result = PreprocessingDataElement(
-                data_element_id=inp.data_element_id,
-                data_element_type=PreprocessingDataType.raw_text_document,
-                data_element_format=PreprocessingDataFormat.txt,
-                data_element_path_or_content=document,
-            )
-            results.append(new_result)
-
-        return PreprocessorResponse(
-            success=True, output_data_type=PreprocessingDataType.raw_text_document, results=results
-        )
-
-    async def preprocess(
-        self,
-        preprocessors: PreprocessorChain,
-        preprocessor_inputs: List[PreprocessingDataElement],
-    ) -> PreprocessorResponse:
-        return await self.do_preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
-
-    @staticmethod
-    def _resolve_input_type(preprocessor_input: PreprocessingDataElement) -> PreprocessingDataType:
-        if preprocessor_input.data_element_type is not None:
-            return preprocessor_input.data_element_type
-
-        if isinstance(preprocessor_input.data_element_path_or_content, URL):
-            return PreprocessingDataType.document_uri
-        if InclineBasicPreprocessorImpl.URL_VALIDATION_PATTERN.match(
-            str(preprocessor_input.data_element_path_or_content)
-        ):
-            return PreprocessingDataType.document_uri
-        if preprocessor_input.data_element_format == PreprocessingDataFormat.pdf:
-            return PreprocessingDataType.binary_document
-
-        return PreprocessingDataType.raw_text_document
-
-    @staticmethod
-    async def _fetch_document(preprocessor_input: PreprocessingDataElement) -> Any:
-        if isinstance(preprocessor_input.data_element_path_or_content, str):
-            url = preprocessor_input.data_element_path_or_content
-            if not InclineBasicPreprocessorImpl.URL_VALIDATION_PATTERN.match(url):
-                log.error(f"Unexpected URL: {url}")
-                return None
-        elif isinstance(preprocessor_input.data_element_path_or_content, URL):
-            url = preprocessor_input.data_element_path_or_content.uri
-        else:
-            log.error(
-                f"Unexpected type {type(preprocessor_input.data_element_path_or_content)} for input {preprocessor_input.data_element_path_or_content}, skipping this input."
-            )
-            return None
-
-        if url.startswith("data:"):
-            return content_from_data(url)
-
-        async with httpx.AsyncClient() as client:
-            r = await client.get(url)
-
-        return r.content if preprocessor_input.data_element_format == PreprocessingDataFormat.pdf else r.text

llama_stack/providers/inline/preprocessing/basic/config.py

@@ -1,9 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-from pydantic import BaseModel
-
-
-class InlineBasicPreprocessorConfig(BaseModel): ...

llama_stack/providers/inline/preprocessing/simple_chunking/__init__.py

@@ -1,18 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-from .config import InclineSimpleChunkingConfig
-
-
-async def get_provider_impl(
-    config: InclineSimpleChunkingConfig,
-    _deps,
-):
-    from .simple_chunking import InclineSimpleChunkingImpl
-
-    impl = InclineSimpleChunkingImpl(config)
-    await impl.initialize()
-    return impl

llama_stack/providers/inline/preprocessing/simple_chunking/config.py

@@ -1,11 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-from pydantic import BaseModel
-
-
-class InclineSimpleChunkingConfig(BaseModel):
-    chunk_size_in_tokens: int = 512
-    chunk_overlap_ratio: int = 4

llama_stack/providers/inline/preprocessing/simple_chunking/simple_chunking.py

@@ -1,116 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-import logging
-from enum import Enum
-from typing import List, Optional, Tuple
-
-from llama_models.llama3.api import Tokenizer
-
-from llama_stack.apis.preprocessing import (
-    Preprocessing,
-    PreprocessingDataElement,
-    PreprocessingDataFormat,
-    PreprocessingDataType,
-    Preprocessor,
-    PreprocessorChain,
-    PreprocessorOptions,
-    PreprocessorResponse,
-)
-from llama_stack.apis.vector_io import Chunk
-from llama_stack.providers.datatypes import PreprocessorsProtocolPrivate
-from llama_stack.providers.inline.preprocessing.simple_chunking import InclineSimpleChunkingConfig
-
-log = logging.getLogger(__name__)
-
-
-class SimpleChunkingOptions(Enum):
-    chunk_size_in_tokens = "chunk_size_in_tokens"
-    chunk_overlap_ratio = "chunk_overlap_ratio"
-
-
-class InclineSimpleChunkingImpl(Preprocessing, PreprocessorsProtocolPrivate):
-    # this preprocessor receives plain text and returns chunks
-    input_types = [PreprocessingDataType.raw_text_document]
-    output_types = [PreprocessingDataType.chunks]
-
-    preprocessor_store = None
-
-    def __init__(self, config: InclineSimpleChunkingConfig) -> None:
-        self.config = config
-
-    async def initialize(self) -> None: ...
-
-    async def shutdown(self) -> None: ...
-
-    async def register_preprocessor(self, preprocessor: Preprocessor) -> None: ...
-
-    async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
-
-    async def do_preprocess(
-        self,
-        preprocessor_id: str,
-        preprocessor_inputs: List[PreprocessingDataElement],
-        options: Optional[PreprocessorOptions] = None,
-    ) -> PreprocessorResponse:
-        chunks = []
-
-        window_len, overlap_len = self._resolve_chunk_size_params(options)
-
-        for inp in preprocessor_inputs:
-            new_chunks = self.make_overlapped_chunks(
-                inp.data_element_id, str(inp.data_element_path_or_content), window_len, overlap_len
-            )
-            for i, chunk in enumerate(new_chunks):
-                new_chunk_data_element = PreprocessingDataElement(
-                    data_element_id=f"{inp.data_element_id}_chunk_{i}",
-                    data_element_type=PreprocessingDataType.chunks,
-                    data_element_format=PreprocessingDataFormat.txt,
-                    data_element_path_or_content=chunk,
-                )
-                chunks.append(new_chunk_data_element)
-
-        return PreprocessorResponse(success=True, output_data_type=PreprocessingDataType.chunks, results=chunks)
-
-    async def preprocess(
-        self,
-        preprocessors: PreprocessorChain,
-        preprocessor_inputs: List[PreprocessingDataElement],
-    ) -> PreprocessorResponse:
-        return await self.do_preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
-
-    def _resolve_chunk_size_params(self, options: PreprocessorOptions | None) -> Tuple[int, int]:
-        window_len = (options or {}).get(
-            str(SimpleChunkingOptions.chunk_size_in_tokens), self.config.chunk_size_in_tokens
-        )
-
-        chunk_overlap_ratio = (options or {}).get(
-            str(SimpleChunkingOptions.chunk_overlap_ratio), self.config.chunk_overlap_ratio
-        )
-        overlap_len = window_len // chunk_overlap_ratio
-
-        return window_len, overlap_len
-
-    @staticmethod
-    def make_overlapped_chunks(document_id: str, text: str, window_len: int, overlap_len: int) -> List[Chunk]:
-        tokenizer = Tokenizer.get_instance()
-        tokens = tokenizer.encode(text, bos=False, eos=False)
-
-        chunks = []
-        for i in range(0, len(tokens), window_len - overlap_len):
-            toks = tokens[i : i + window_len]
-            chunk = tokenizer.decode(toks)
-            # chunk is a string
-            chunks.append(
-                Chunk(
-                    content=chunk,
-                    metadata={
-                        "token_count": len(toks),
-                        "document_id": document_id,
-                    },
-                )
-            )
-
-        return chunks

llama_stack/providers/inline/tool_runtime/rag/__init__.py

@@ -14,6 +14,6 @@ from .config import RagToolRuntimeConfig
 async def get_provider_impl(config: RagToolRuntimeConfig, deps: Dict[Api, Any]):
     from .memory import MemoryToolRuntimeImpl
 
-    impl = MemoryToolRuntimeImpl(config, deps[Api.vector_io], deps[Api.inference], deps[Api.preprocessing])
+    impl = MemoryToolRuntimeImpl(config, deps[Api.vector_io], deps[Api.inference])
     await impl.initialize()
     return impl

llama_stack/providers/inline/tool_runtime/rag/memory.py

@@ -19,14 +19,6 @@ from llama_stack.apis.common.content_types import (
     TextContentItem,
 )
 from llama_stack.apis.inference import Inference
-from llama_stack.apis.preprocessing import (
-    Preprocessing,
-    PreprocessingDataElement,
-    PreprocessingDataFormat,
-    PreprocessingDataType,
-    PreprocessorChain,
-    PreprocessorChainElement,
-)
 from llama_stack.apis.tools import (
     ListToolDefsResponse,
     RAGDocument,
@@ -41,6 +33,10 @@ from llama_stack.apis.tools import (
 )
 from llama_stack.apis.vector_io import QueryChunksResponse, VectorIO
 from llama_stack.providers.datatypes import ToolsProtocolPrivate
+from llama_stack.providers.utils.memory.vector_store import (
+    content_from_doc,
+    make_overlapped_chunks,
+)
 
 from .config import RagToolRuntimeConfig
 from .context_retriever import generate_rag_query
@@ -53,22 +49,15 @@ def make_random_string(length: int = 8):
 
 
 class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
-    DEFAULT_PREPROCESSING_CHAIN = [
-        PreprocessorChainElement(preprocessor_id="builtin::basic"),
-        PreprocessorChainElement(preprocessor_id="builtin::chunking"),
-    ]
-
     def __init__(
         self,
         config: RagToolRuntimeConfig,
         vector_io_api: VectorIO,
         inference_api: Inference,
-        preprocessing_api: Preprocessing,
     ):
         self.config = config
         self.vector_io_api = vector_io_api
         self.inference_api = inference_api
-        self.preprocessing_api = preprocessing_api
 
     async def initialize(self):
         pass
@@ -87,32 +76,24 @@ class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
         documents: List[RAGDocument],
         vector_db_id: str,
         chunk_size_in_tokens: int = 512,
-        preprocessor_chain: Optional[PreprocessorChain] = None,
     ) -> None:
-        preprocessor_inputs = [self._rag_document_to_preprocessor_input(d) for d in documents]
-        preprocessor_response = await self.preprocessing_api.preprocess(
-            preprocessors=preprocessor_chain or self.DEFAULT_PREPROCESSING_CHAIN,
-            preprocessor_inputs=preprocessor_inputs,
-        )
-
-        if not preprocessor_response.success:
-            log.error("Preprocessor chain returned an error")
-            return
-
-        if preprocessor_response.output_data_type != PreprocessingDataType.chunks:
-            log.error(
-                f"Preprocessor chain returned {preprocessor_response.output_data_type} instead of {PreprocessingDataType.chunks}"
+        chunks = []
+        for doc in documents:
+            content = await content_from_doc(doc)
+            chunks.extend(
+                make_overlapped_chunks(
+                    doc.document_id,
+                    content,
+                    chunk_size_in_tokens,
+                    chunk_size_in_tokens // 4,
+                )
             )
-            return
 
-        chunks = preprocessor_response.results
         if not chunks:
-            log.error("No chunks returned by the preprocessor chain")
             return
 
-        actual_chunks = [chunk.data_element_path_or_content for chunk in chunks]
         await self.vector_io_api.insert_chunks(
-            chunks=actual_chunks,  # type: ignore
+            chunks=chunks,
             vector_db_id=vector_db_id,
         )
 
@@ -226,16 +207,3 @@ class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
             content=result.content,
             metadata=result.metadata,
         )
-
-    @staticmethod
-    def _rag_document_to_preprocessor_input(document: RAGDocument) -> PreprocessingDataElement:
-        if document.mime_type == "application/pdf":
-            data_element_format = PreprocessingDataFormat.pdf
-        else:
-            data_element_format = None
-
-        return PreprocessingDataElement(
-            data_element_id=document.document_id,
-            data_element_format=data_element_format,
-            data_element_path_or_content=document.content,
-        )

llama_stack/providers/registry/preprocessing.py

@@ -7,28 +7,9 @@
 from typing import List
 
 from llama_stack.providers.datatypes import (
-    Api,
-    InlineProviderSpec,
     ProviderSpec,
 )
 
 
 def available_providers() -> List[ProviderSpec]:
-    return [
-        InlineProviderSpec(
-            api=Api.preprocessing,
-            provider_type="inline::basic",
-            pip_packages=["httpx", "pypdf"],
-            module="llama_stack.providers.inline.preprocessing.basic",
-            config_class="llama_stack.providers.inline.preprocessing.basic.InlineBasicPreprocessorConfig",
-            api_dependencies=[],
-        ),
-        InlineProviderSpec(
-            api=Api.preprocessing,
-            provider_type="inline::simple_chunking",
-            pip_packages=[],
-            module="llama_stack.providers.inline.preprocessing.simple_chunking",
-            config_class="llama_stack.providers.inline.preprocessing.simple_chunking.InclineSimpleChunkingConfig",
-            api_dependencies=[],
-        ),
-    ]
+    return []

llama_stack/providers/registry/tool_runtime.py

@@ -34,7 +34,7 @@ def available_providers() -> List[ProviderSpec]:
             ],
             module="llama_stack.providers.inline.tool_runtime.rag",
             config_class="llama_stack.providers.inline.tool_runtime.rag.config.RagToolRuntimeConfig",
-            api_dependencies=[Api.vector_io, Api.inference, Api.preprocessing],
+            api_dependencies=[Api.vector_io, Api.inference],
         ),
         InlineProviderSpec(
             api=Api.tool_runtime,

llama_stack/templates/bedrock/bedrock.py

@@ -6,7 +6,6 @@
 
 from pathlib import Path
 
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import Provider, ToolGroupInput
 from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
 from llama_stack.providers.remote.inference.bedrock.models import MODEL_ENTRIES
@@ -34,7 +33,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "bedrock"
     vector_io_provider = Provider(
@@ -63,17 +61,6 @@ def get_distribution_template() -> DistributionTemplate:
         ),
     ]
 
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
-
     return DistributionTemplate(
         name=name,
         distro_type="self_hosted",
@@ -89,7 +76,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models,
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/bedrock/build.yaml

@@ -29,7 +29,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/bedrock/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -97,13 +96,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/bedrock}/registry.db
@@ -150,10 +142,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/cerebras/build.yaml

@@ -29,7 +29,4 @@ distribution_spec:
     - remote::tavily-search
     - inline::code-interpreter
     - inline::rag-runtime
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/cerebras/cerebras.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput
 from llama_stack.providers.inline.inference.sentence_transformers import (
     SentenceTransformersInferenceConfig,
@@ -38,7 +37,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::code-interpreter",
             "inline::rag-runtime",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
 
     name = "cerebras"
@@ -84,16 +82,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name="cerebras",
@@ -112,7 +100,6 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=default_models + [embedding_model],
                 default_shields=[],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/cerebras/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -100,13 +99,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/cerebras}/registry.db
@@ -148,10 +140,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/ci-tests/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/ci-tests/ci_tests.py

@@ -6,7 +6,6 @@
 
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -45,7 +44,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "ci-tests"
     inference_provider = Provider(
@@ -78,16 +76,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
     available_models = {
         "fireworks": MODEL_ENTRIES,
     }
@@ -117,7 +105,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models + [embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
                 default_shields=[ShieldInput(shield_id="meta-llama/Llama-Guard-3-8B")],
             ),
         },

llama_stack/templates/ci-tests/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -100,13 +99,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/ci-tests}/registry.db
@@ -226,10 +218,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/dell/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - remote::tavily-search
     - inline::code-interpreter
     - inline::rag-runtime
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/dell/dell.py

@@ -5,7 +5,6 @@
 # the root directory of this source tree.
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -34,7 +33,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::code-interpreter",
             "inline::rag-runtime",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "dell"
     inference_provider = Provider(
@@ -94,16 +92,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -119,7 +107,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -133,7 +120,6 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=[inference_model, safety_model, embedding_model],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/dell/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -100,13 +99,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/dell}/registry.db
@@ -137,10 +129,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/dell/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -96,13 +95,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/dell}/registry.db
@@ -128,10 +120,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/dev/build.yaml

@@ -34,7 +34,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/dev/dev.py

@@ -7,7 +7,6 @@
 from typing import List, Tuple
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -112,7 +111,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "dev"
 
@@ -157,16 +155,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
     embedding_model = ModelInput(
         model_id="all-MiniLM-L6-v2",
         provider_id=embedding_provider.provider_id,
@@ -193,7 +181,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models + [embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
                 default_shields=[ShieldInput(shield_id="meta-llama/Llama-Guard-3-8B")],
             ),
         },

llama_stack/templates/dev/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -129,13 +128,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/dev}/registry.db
@@ -382,10 +374,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/fireworks/build.yaml

@@ -31,7 +31,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/fireworks/fireworks.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -45,7 +44,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
 
     name = "fireworks"
@@ -97,16 +95,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -125,7 +113,6 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=default_models + [embedding_model],
                 default_shields=[ShieldInput(shield_id="meta-llama/Llama-Guard-3-8B")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -171,7 +158,6 @@ def get_distribution_template() -> DistributionTemplate:
                     ),
                 ],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/fireworks/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -112,13 +111,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/fireworks}/registry.db
@@ -245,10 +237,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/fireworks/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -107,13 +106,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/fireworks}/registry.db
@@ -235,10 +227,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/groq/build.yaml

@@ -26,7 +26,4 @@ distribution_spec:
     - remote::tavily-search
     - inline::code-interpreter
     - inline::rag-runtime
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/groq/groq.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput
 from llama_stack.providers.inline.inference.sentence_transformers import (
     SentenceTransformersInferenceConfig,
@@ -37,7 +36,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::code-interpreter",
             "inline::rag-runtime",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "groq"
 
@@ -79,16 +77,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -105,7 +93,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models + [embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/groq/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -100,13 +99,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/groq}/registry.db
@@ -173,10 +165,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/hf-endpoint/build.yaml

@@ -29,7 +29,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/hf-endpoint/hf_endpoint.py

@@ -5,7 +5,6 @@
 # the root directory of this source tree.
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -37,7 +36,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "hf-endpoint"
     inference_provider = Provider(
@@ -86,16 +84,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -112,7 +100,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -136,7 +123,6 @@ def get_distribution_template() -> DistributionTemplate:
                 ],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/hf-endpoint/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -108,13 +107,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/hf-endpoint}/registry.db
@@ -145,10 +137,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/hf-endpoint/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -103,13 +102,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/hf-endpoint}/registry.db
@@ -135,10 +127,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/hf-serverless/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/hf-serverless/hf_serverless.py

@@ -5,7 +5,6 @@
 # the root directory of this source tree.
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -37,7 +36,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
 
     name = "hf-serverless"
@@ -87,16 +85,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -113,7 +101,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -137,7 +124,6 @@ def get_distribution_template() -> DistributionTemplate:
                 ],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/hf-serverless/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -108,13 +107,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/hf-serverless}/registry.db
@@ -145,10 +137,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/hf-serverless/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -103,13 +102,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/hf-serverless}/registry.db
@@ -135,10 +127,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/meta-reference-gpu/build.yaml

@@ -29,7 +29,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/meta-reference-gpu/meta_reference.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -41,7 +40,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "meta-reference-gpu"
     inference_provider = Provider(
@@ -93,16 +91,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -118,7 +106,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -143,7 +130,6 @@ def get_distribution_template() -> DistributionTemplate:
                 ],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/meta-reference-gpu/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -110,13 +109,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/registry.db
@@ -147,10 +139,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/meta-reference-gpu/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -104,13 +103,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-gpu}/registry.db
@@ -136,10 +128,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/meta-reference-quantized-gpu/build.yaml

@@ -29,7 +29,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/meta-reference-quantized-gpu/meta_reference.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput
 from llama_stack.providers.inline.inference.meta_reference import (
     MetaReferenceQuantizedInferenceConfig,
@@ -36,7 +35,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     default_tool_groups = [
         ToolGroupInput(
@@ -52,16 +50,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
     name = "meta-reference-quantized-gpu"
     inference_provider = Provider(
         provider_id="meta-reference-inference",
@@ -108,7 +96,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/meta-reference-quantized-gpu/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -106,13 +105,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/meta-reference-quantized-gpu}/registry.db
@@ -138,10 +130,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/nvidia/build.yaml

@@ -22,7 +22,4 @@ distribution_spec:
     - inline::basic
     tool_runtime:
     - inline::rag-runtime
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/nvidia/nvidia.py

@@ -6,7 +6,6 @@
 
 from pathlib import Path
 
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import ModelInput, Provider, ShieldInput, ToolGroupInput
 from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig
 from llama_stack.providers.remote.inference.nvidia.models import MODEL_ENTRIES
@@ -26,7 +25,6 @@ def get_distribution_template() -> DistributionTemplate:
         "datasetio": ["inline::localfs"],
         "scoring": ["inline::basic"],
         "tool_runtime": ["inline::rag-runtime"],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
 
     inference_provider = Provider(
@@ -57,16 +55,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="rag-runtime",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     default_models = get_model_registry(available_models)
     return DistributionTemplate(
@@ -84,7 +72,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models,
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -96,7 +83,6 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=[inference_model, safety_model],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}", provider_id="nvidia")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/nvidia/run-with-safety.yaml

@@ -6,7 +6,6 @@ apis:
 - eval
 - inference
 - post_training
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -85,13 +84,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
@@ -114,10 +106,6 @@ benchmarks: []
 tool_groups:
 - toolgroup_id: builtin::rag
   provider_id: rag-runtime
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/nvidia/run.yaml

@@ -6,7 +6,6 @@ apis:
 - eval
 - inference
 - post_training
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -80,13 +79,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/nvidia}/registry.db
@@ -217,10 +209,6 @@ benchmarks: []
 tool_groups:
 - toolgroup_id: builtin::rag
   provider_id: rag-runtime
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/ollama/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - inline::rag-runtime
     - remote::model-context-protocol
     - remote::wolfram-alpha
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/ollama/ollama.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -37,7 +36,6 @@ def get_distribution_template() -> DistributionTemplate:
             "remote::model-context-protocol",
             "remote::wolfram-alpha",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "ollama"
     inference_provider = Provider(
@@ -86,16 +84,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="wolfram-alpha",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -112,7 +100,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -147,7 +134,6 @@ def get_distribution_template() -> DistributionTemplate:
                     ),
                 ],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/ollama/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -105,13 +104,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/ollama}/registry.db
@@ -148,10 +140,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/ollama/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -103,13 +102,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/ollama}/registry.db
@@ -138,10 +130,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/remote-vllm/build.yaml

@@ -31,7 +31,4 @@ distribution_spec:
     - inline::rag-runtime
     - remote::model-context-protocol
     - remote::wolfram-alpha
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/remote-vllm/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -116,13 +115,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/remote-vllm}/registry.db
@@ -155,10 +147,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/remote-vllm/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -109,13 +108,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/remote-vllm}/registry.db
@@ -143,10 +135,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/remote-vllm/vllm.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -40,7 +39,6 @@ def get_distribution_template() -> DistributionTemplate:
             "remote::model-context-protocol",
             "remote::wolfram-alpha",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "remote-vllm"
     inference_provider = Provider(
@@ -95,16 +93,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="wolfram-alpha",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -120,7 +108,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -144,7 +131,6 @@ def get_distribution_template() -> DistributionTemplate:
                 ],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/sambanova/build.yaml

@@ -19,7 +19,4 @@ distribution_spec:
     - remote::tavily-search
     - inline::code-interpreter
     - inline::rag-runtime
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/sambanova/run.yaml

@@ -3,7 +3,6 @@ image_name: sambanova
 apis:
 - agents
 - inference
-- preprocessing
 - safety
 - telemetry
 - tool_runtime
@@ -72,13 +71,6 @@ providers:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/sambanova}/registry.db
@@ -186,10 +178,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/sambanova/sambanova.py

@@ -6,7 +6,6 @@
 
 from pathlib import Path
 
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import Provider, ShieldInput, ToolGroupInput
 from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
 from llama_stack.providers.remote.inference.sambanova import SambaNovaImplConfig
@@ -35,7 +34,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::code-interpreter",
             "inline::rag-runtime",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "sambanova"
 
@@ -87,16 +85,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -115,7 +103,6 @@ def get_distribution_template() -> DistributionTemplate:
                 default_models=default_models,
                 default_shields=[ShieldInput(shield_id="meta-llama/Llama-Guard-3-8B")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/tgi/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/tgi/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -103,13 +102,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/tgi}/registry.db
@@ -135,10 +127,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/tgi/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -102,13 +101,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/tgi}/registry.db
@@ -134,10 +126,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/tgi/tgi.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -39,7 +38,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "tgi"
     inference_provider = Provider(
@@ -90,16 +88,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -116,7 +104,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
             "run-with-safety.yaml": RunConfigSettings(
                 provider_overrides={
@@ -138,7 +125,6 @@ def get_distribution_template() -> DistributionTemplate:
                 ],
                 default_shields=[ShieldInput(shield_id="${env.SAFETY_MODEL}")],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/together/build.yaml

@@ -31,7 +31,4 @@ distribution_spec:
     - inline::rag-runtime
     - remote::model-context-protocol
     - remote::wolfram-alpha
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/together/run-with-safety.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -112,13 +111,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/together}/registry.db
@@ -252,10 +244,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/together/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -107,13 +106,6 @@ providers:
     provider_type: remote::wolfram-alpha
     config:
       api_key: ${env.WOLFRAM_ALPHA_API_KEY:}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/together}/registry.db
@@ -242,10 +234,6 @@ tool_groups:
   provider_id: code-interpreter
 - toolgroup_id: builtin::wolfram_alpha
   provider_id: wolfram-alpha
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/together/together.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import (
     ModelInput,
     Provider,
@@ -45,7 +44,6 @@ def get_distribution_template() -> DistributionTemplate:
             "remote::model-context-protocol",
             "remote::wolfram-alpha",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
     name = "together"
     inference_provider = Provider(
@@ -85,16 +83,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="wolfram-alpha",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
     embedding_model = ModelInput(
         model_id="all-MiniLM-L6-v2",
         provider_id="sentence-transformers",
@@ -120,7 +108,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=default_models + [embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
                 default_shields=[ShieldInput(shield_id="meta-llama/Llama-Guard-3-8B")],
             ),
             "run-with-safety.yaml": RunConfigSettings(
@@ -167,7 +154,6 @@ def get_distribution_template() -> DistributionTemplate:
                     ),
                 ],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={

llama_stack/templates/vllm-gpu/build.yaml

@@ -30,7 +30,4 @@ distribution_spec:
     - inline::code-interpreter
     - inline::rag-runtime
     - remote::model-context-protocol
-    preprocessing:
-    - inline::basic
-    - inline::simple_chunking
 image_type: conda

llama_stack/templates/vllm-gpu/run.yaml

@@ -5,7 +5,6 @@ apis:
 - datasetio
 - eval
 - inference
-- preprocessing
 - safety
 - scoring
 - telemetry
@@ -107,13 +106,6 @@ providers:
   - provider_id: model-context-protocol
     provider_type: remote::model-context-protocol
     config: {}
-  preprocessing:
-  - provider_id: basic
-    provider_type: inline::basic
-    config: {}
-  - provider_id: simple_chunking
-    provider_type: inline::simple_chunking
-    config: {}
 metadata_store:
   type: sqlite
   db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/vllm-gpu}/registry.db
@@ -139,10 +131,6 @@ tool_groups:
   provider_id: rag-runtime
 - toolgroup_id: builtin::code_interpreter
   provider_id: code-interpreter
-preprocessors:
-- preprocessor_id: builtin::basic
-  provider_id: basic
-- preprocessor_id: builtin::chunking
-  provider_id: simple_chunking
+preprocessors: []
 server:
   port: 8321

llama_stack/templates/vllm-gpu/vllm.py

@@ -5,7 +5,6 @@
 # the root directory of this source tree.
 
 from llama_stack.apis.models.models import ModelType
-from llama_stack.apis.preprocessors.preprocessors import PreprocessorInput
 from llama_stack.distribution.datatypes import ModelInput, Provider
 from llama_stack.providers.inline.inference.sentence_transformers import (
     SentenceTransformersInferenceConfig,
@@ -36,7 +35,6 @@ def get_distribution_template() -> DistributionTemplate:
             "inline::rag-runtime",
             "remote::model-context-protocol",
         ],
-        "preprocessing": ["inline::basic", "inline::simple_chunking"],
     }
 
     name = "vllm-gpu"
@@ -82,16 +80,6 @@ def get_distribution_template() -> DistributionTemplate:
             provider_id="code-interpreter",
         ),
     ]
-    default_preprocessors = [
-        PreprocessorInput(
-            preprocessor_id="builtin::basic",
-            provider_id="basic",
-        ),
-        PreprocessorInput(
-            preprocessor_id="builtin::chunking",
-            provider_id="simple_chunking",
-        ),
-    ]
 
     return DistributionTemplate(
         name=name,
@@ -108,7 +96,6 @@ def get_distribution_template() -> DistributionTemplate:
                 },
                 default_models=[inference_model, embedding_model],
                 default_tool_groups=default_tool_groups,
-                default_preprocessors=default_preprocessors,
             ),
         },
         run_config_env_vars={