diff --git a/.github/workflows/README.md b/.github/workflows/README.md
index 059bb873f..4ba67c148 100644
--- a/.github/workflows/README.md
+++ b/.github/workflows/README.md
@@ -13,6 +13,10 @@ Llama Stack uses GitHub Actions for Continuous Integration (CI). Below is a tabl
 | Vector IO Integration Tests | [integration-vector-io-tests.yml](integration-vector-io-tests.yml) | Run the integration test suite with various VectorIO providers |
 | Pre-commit | [pre-commit.yml](pre-commit.yml) | Run pre-commit checks |
 | Test Llama Stack Build | [providers-build.yml](providers-build.yml) | Test llama stack build |
+<<<<<<< Updated upstream
+=======
+| Test Llama Stack Show | [providers-show.yml](providers-show.yml) | Test llama stack Show |
+>>>>>>> Stashed changes
 | Python Package Build Test | [python-build-test.yml](python-build-test.yml) | Test building the llama-stack PyPI project |
 | Integration Tests (Record) | [record-integration-tests.yml](record-integration-tests.yml) | Run the integration test suite from tests/integration |
 | Check semantic PR titles | [semantic-pr.yml](semantic-pr.yml) | Ensure that PR titles follow the conventional commit spec |
diff --git a/docs/source/providers/post_training/inline_torchtune-cpu.md b/docs/source/providers/post_training/inline_torchtune-cpu.md
index 7204e56e8..78e4da379 100644
--- a/docs/source/providers/post_training/inline_torchtune-cpu.md
+++ b/docs/source/providers/post_training/inline_torchtune-cpu.md
@@ -2,7 +2,7 @@
 
 ## Description
 
-TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework.
+TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework (CPU).
 
 ## Configuration
 
diff --git a/docs/source/providers/post_training/inline_torchtune-gpu.md b/docs/source/providers/post_training/inline_torchtune-gpu.md
index 98b94f6f6..9ac28b503 100644
--- a/docs/source/providers/post_training/inline_torchtune-gpu.md
+++ b/docs/source/providers/post_training/inline_torchtune-gpu.md
@@ -2,7 +2,7 @@
 
 ## Description
 
-TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework.
+TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework (GPU).
 
 ## Configuration
 
diff --git a/llama_stack/providers/datatypes.py b/llama_stack/providers/datatypes.py
index c8ff9cecb..77df6f4b9 100644
--- a/llama_stack/providers/datatypes.py
+++ b/llama_stack/providers/datatypes.py
@@ -141,7 +141,10 @@ class ProviderSpec(BaseModel):
     )
 
     is_external: bool = Field(default=False, description="Notes whether this provider is an external provider.")
-
+    package_extras: list[str] = Field(
+        default_factory=list,
+        description="Optional package extras to install when using pyproject.toml files (e.g., ['cpu', 'gpu'])",
+    )
     # used internally by the resolver; this is a hack for now
     deps__: list[str] = Field(default_factory=list)
 
diff --git a/llama_stack/providers/inline/agents/meta_reference/pyproject.toml b/llama_stack/providers/inline/agents/meta_reference/pyproject.toml
new file mode 100644
index 000000000..c32655b52
--- /dev/null
+++ b/llama_stack/providers/inline/agents/meta_reference/pyproject.toml
@@ -0,0 +1,28 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-agents-meta-reference"
+version = "0.1.0"
+description = "Meta's reference implementation of an agent system that can use tools, access vector databases, and perform complex reasoning tasks"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "matplotlib",
+    "pillow",
+    "pandas",
+    "scikit-learn",
+    "mcp>=1.8.1",
+    "aiosqlite",
+    "psycopg2-binary",
+    "redis",
+    "pymongo"
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/batches/reference/pyproject.toml b/llama_stack/providers/inline/batches/reference/pyproject.toml
new file mode 100644
index 000000000..13cadec1e
--- /dev/null
+++ b/llama_stack/providers/inline/batches/reference/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-batches-meta-reference"
+version = "0.1.0"
+description = "Meta's reference implementation of batches API with KVStore persistence."
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "openai"
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/datasetio/localfs/pyproject.toml b/llama_stack/providers/inline/datasetio/localfs/pyproject.toml
new file mode 100644
index 000000000..2e0c33eac
--- /dev/null
+++ b/llama_stack/providers/inline/datasetio/localfs/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-datasetio-localfs"
+version = "0.1.0"
+description = "Local filesystem-based dataset I/O provider for reading and writing datasets to local storage"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "pandas",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/eval/meta_reference/pyproject.toml b/llama_stack/providers/inline/eval/meta_reference/pyproject.toml
new file mode 100644
index 000000000..d0a0a5e38
--- /dev/null
+++ b/llama_stack/providers/inline/eval/meta_reference/pyproject.toml
@@ -0,0 +1,24 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-eval-meta-reference"
+version = "0.1.0"
+description = "Meta's reference implementation of evaluation tasks with support for multiple languages and evaluation metrics"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "tree_sitter",
+    "pythainlp",
+    "langdetect",
+    "emoji",
+    "nltk",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/files/localfs/pyproject.toml b/llama_stack/providers/inline/files/localfs/pyproject.toml
new file mode 100644
index 000000000..814d2e71d
--- /dev/null
+++ b/llama_stack/providers/inline/files/localfs/pyproject.toml
@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-files-localfs"
+version = "0.1.0"
+description = "Local filesystem-based file storage provider for managing files and documents locally"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = []
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/inference/meta_reference/pyproject.toml b/llama_stack/providers/inline/inference/meta_reference/pyproject.toml
new file mode 100644
index 000000000..9618e8f0b
--- /dev/null
+++ b/llama_stack/providers/inline/inference/meta_reference/pyproject.toml
@@ -0,0 +1,29 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-meta-reference"
+version = "0.1.0"
+description = "Meta's reference implementation of inference with support for various model formats and optimization techniques"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "accelerate",
+    "fairscale",
+    "torch",
+    "torchvision",
+    "transformers",
+    "zmq",
+    "lm-format-enforcer",
+    "sentence-transformers",
+    "torchao==0.8.0",
+    "fbgemm-gpu-genai==1.1.2",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/inference/sentence_transformers/pyproject.toml b/llama_stack/providers/inline/inference/sentence_transformers/pyproject.toml
new file mode 100644
index 000000000..2221cb6cb
--- /dev/null
+++ b/llama_stack/providers/inline/inference/sentence_transformers/pyproject.toml
@@ -0,0 +1,22 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-sentence-transformers"
+version = "0.1.0"
+description = "Sentence Transformers inference provider for text embeddings and similarity search"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "torch",
+    "torchvision",
+    "sentence-transformers",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/post_training/huggingface/pyproject.toml b/llama_stack/providers/inline/post_training/huggingface/pyproject.toml
new file mode 100644
index 000000000..54f0980f5
--- /dev/null
+++ b/llama_stack/providers/inline/post_training/huggingface/pyproject.toml
@@ -0,0 +1,24 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-post-training-huggingface"
+version = "0.1.0"
+description = "HuggingFace-based post-training provider for fine-tuning models using the HuggingFace ecosystem"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "trl",
+    "transformers",
+    "peft",
+    "datasets>=4.0.0",
+    "torch",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/post_training/torchtune/pyproject.toml b/llama_stack/providers/inline/post_training/torchtune/pyproject.toml
new file mode 100644
index 000000000..e6f475b6e
--- /dev/null
+++ b/llama_stack/providers/inline/post_training/torchtune/pyproject.toml
@@ -0,0 +1,30 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-post-training-torchtune"
+version = "0.1.0"
+description = "TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "torchtune>=0.5.0",
+    "torchao>=0.12.0",
+    "numpy",
+]
+
+[project.optional-dependencies]
+cpu = [
+    "torch --extra-index-url https://download.pytorch.org/whl/cpu",
+]
+gpu = [
+    "torch",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/safety/code_scanner/pyproject.toml b/llama_stack/providers/inline/safety/code_scanner/pyproject.toml
new file mode 100644
index 000000000..0cd40a88b
--- /dev/null
+++ b/llama_stack/providers/inline/safety/code_scanner/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-code-scanner"
+version = "0.1.0"
+description = "Code Scanner safety provider for detecting security vulnerabilities and unsafe code patterns"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "codeshield",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/safety/llama_guard/pyproject.toml b/llama_stack/providers/inline/safety/llama_guard/pyproject.toml
new file mode 100644
index 000000000..71df3238c
--- /dev/null
+++ b/llama_stack/providers/inline/safety/llama_guard/pyproject.toml
@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-llama-guard"
+version = "0.1.0"
+description = "Llama Guard safety provider for content moderation and safety filtering using Meta's Llama Guard model"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = []
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/safety/prompt_guard/pyproject.toml b/llama_stack/providers/inline/safety/prompt_guard/pyproject.toml
new file mode 100644
index 000000000..cee17b302
--- /dev/null
+++ b/llama_stack/providers/inline/safety/prompt_guard/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-prompt-guard"
+version = "0.1.0"
+description = "Prompt Guard safety provider for detecting and filtering unsafe prompts and content"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "transformers[accelerate]",
+    "torch",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/scoring/basic/pyproject.toml b/llama_stack/providers/inline/scoring/basic/pyproject.toml
new file mode 100644
index 000000000..1fc86c36c
--- /dev/null
+++ b/llama_stack/providers/inline/scoring/basic/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-scoring-basic"
+version = "0.1.0"
+description = "Basic scoring provider for simple evaluation metrics and scoring functions"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/scoring/braintrust/pyproject.toml b/llama_stack/providers/inline/scoring/braintrust/pyproject.toml
new file mode 100644
index 000000000..c5d0780c8
--- /dev/null
+++ b/llama_stack/providers/inline/scoring/braintrust/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-scoring-braintrust"
+version = "0.1.0"
+description = "Braintrust scoring provider for evaluation and scoring using the Braintrust platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "autoevals",
+    "openai",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/scoring/llm_as_judge/pyproject.toml b/llama_stack/providers/inline/scoring/llm_as_judge/pyproject.toml
new file mode 100644
index 000000000..46cdc96ea
--- /dev/null
+++ b/llama_stack/providers/inline/scoring/llm_as_judge/pyproject.toml
@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-scoring-llm-as-judge"
+version = "0.1.0"
+description = "LLM-as-judge scoring provider that uses language models to evaluate and score responses"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = []
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/telemetry/meta_reference/pyproject.toml b/llama_stack/providers/inline/telemetry/meta_reference/pyproject.toml
new file mode 100644
index 000000000..23d41f7ba
--- /dev/null
+++ b/llama_stack/providers/inline/telemetry/meta_reference/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-telemetry-meta-reference"
+version = "0.1.0"
+description = "Meta's reference implementation of telemetry and observability using OpenTelemetry"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "opentelemetry-sdk",
+    "opentelemetry-exporter-otlp-proto-http",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/tool_runtime/rag/pyproject.toml b/llama_stack/providers/inline/tool_runtime/rag/pyproject.toml
new file mode 100644
index 000000000..53510b2d4
--- /dev/null
+++ b/llama_stack/providers/inline/tool_runtime/rag/pyproject.toml
@@ -0,0 +1,28 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-rag"
+version = "0.1.0"
+description = "RAG (Retrieval-Augmented Generation) tool runtime for document ingestion, chunking, and semantic search"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "chardet",
+    "pypdf",
+    "tqdm",
+    "numpy",
+    "scikit-learn",
+    "scipy",
+    "nltk",
+    "sentencepiece",
+    "transformers",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/vector_io/chroma/pyproject.toml b/llama_stack/providers/inline/vector_io/chroma/pyproject.toml
new file mode 100644
index 000000000..dd33ed593
--- /dev/null
+++ b/llama_stack/providers/inline/vector_io/chroma/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-chroma"
+version = "0.1.0"
+description = "Chroma inline vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "chromadb",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/vector_io/faiss/pyproject.toml b/llama_stack/providers/inline/vector_io/faiss/pyproject.toml
new file mode 100644
index 000000000..712d70ba2
--- /dev/null
+++ b/llama_stack/providers/inline/vector_io/faiss/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-faiss"
+version = "0.1.0"
+description = "Faiss inline vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "faiss-cpu",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/vector_io/milvus/pyproject.toml b/llama_stack/providers/inline/vector_io/milvus/pyproject.toml
new file mode 100644
index 000000000..3bde5a531
--- /dev/null
+++ b/llama_stack/providers/inline/vector_io/milvus/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-milvus"
+version = "0.1.0"
+description = "Milvus inline vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "pymilvus[milvus-lite]>=2.4.10",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/vector_io/qdrant/pyproject.toml b/llama_stack/providers/inline/vector_io/qdrant/pyproject.toml
new file mode 100644
index 000000000..c2b7ee82d
--- /dev/null
+++ b/llama_stack/providers/inline/vector_io/qdrant/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-qdrant"
+version = "0.1.0"
+description = "Qdrant inline vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "qdrant-client",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/inline/vector_io/sqlite_vec/pyproject.toml b/llama_stack/providers/inline/vector_io/sqlite_vec/pyproject.toml
new file mode 100644
index 000000000..37cb0bf67
--- /dev/null
+++ b/llama_stack/providers/inline/vector_io/sqlite_vec/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-sqlite-vec"
+version = "0.1.0"
+description = "SQLite-Vec inline vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "sqlite-vec",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/registry/agents.py b/llama_stack/providers/registry/agents.py
index 57110d129..f367232ef 100644
--- a/llama_stack/providers/registry/agents.py
+++ b/llama_stack/providers/registry/agents.py
@@ -10,7 +10,6 @@ from llama_stack.providers.datatypes import (
     InlineProviderSpec,
     ProviderSpec,
 )
-from llama_stack.providers.utils.kvstore import kvstore_dependencies
 
 
 def available_providers() -> list[ProviderSpec]:
@@ -18,14 +17,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.agents,
             provider_type="inline::meta-reference",
-            pip_packages=[
-                "matplotlib",
-                "pillow",
-                "pandas",
-                "scikit-learn",
-                "mcp>=1.8.1",
-            ]
-            + kvstore_dependencies(),  # TODO make this dynamic based on the kvstore config
             module="llama_stack.providers.inline.agents.meta_reference",
             config_class="llama_stack.providers.inline.agents.meta_reference.MetaReferenceAgentsImplConfig",
             api_dependencies=[
diff --git a/llama_stack/providers/registry/batches.py b/llama_stack/providers/registry/batches.py
index a07942486..7c0f5d75b 100644
--- a/llama_stack/providers/registry/batches.py
+++ b/llama_stack/providers/registry/batches.py
@@ -13,7 +13,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.batches,
             provider_type="inline::reference",
-            pip_packages=[],
             module="llama_stack.providers.inline.batches.reference",
             config_class="llama_stack.providers.inline.batches.reference.config.ReferenceBatchesImplConfig",
             api_dependencies=[
diff --git a/llama_stack/providers/registry/datasetio.py b/llama_stack/providers/registry/datasetio.py
index a9feb0bac..7e56ff778 100644
--- a/llama_stack/providers/registry/datasetio.py
+++ b/llama_stack/providers/registry/datasetio.py
@@ -18,7 +18,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.datasetio,
             provider_type="inline::localfs",
-            pip_packages=["pandas"],
             module="llama_stack.providers.inline.datasetio.localfs",
             config_class="llama_stack.providers.inline.datasetio.localfs.LocalFSDatasetIOConfig",
             api_dependencies=[],
@@ -28,9 +27,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.datasetio,
             adapter_type="huggingface",
             provider_type="remote::huggingface",
-            pip_packages=[
-                "datasets>=4.0.0",
-            ],
             module="llama_stack.providers.remote.datasetio.huggingface",
             config_class="llama_stack.providers.remote.datasetio.huggingface.HuggingfaceDatasetIOConfig",
             description="HuggingFace datasets provider for accessing and managing datasets from the HuggingFace Hub.",
@@ -41,9 +37,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::nvidia",
             module="llama_stack.providers.remote.datasetio.nvidia",
             config_class="llama_stack.providers.remote.datasetio.nvidia.NvidiaDatasetIOConfig",
-            pip_packages=[
-                "datasets>=4.0.0",
-            ],
             description="NVIDIA's dataset I/O provider for accessing datasets from NVIDIA's data platform.",
         ),
     ]
diff --git a/llama_stack/providers/registry/eval.py b/llama_stack/providers/registry/eval.py
index 4ef0bb41f..e0074a574 100644
--- a/llama_stack/providers/registry/eval.py
+++ b/llama_stack/providers/registry/eval.py
@@ -13,7 +13,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.eval,
             provider_type="inline::meta-reference",
-            pip_packages=["tree_sitter", "pythainlp", "langdetect", "emoji", "nltk"],
             module="llama_stack.providers.inline.eval.meta_reference",
             config_class="llama_stack.providers.inline.eval.meta_reference.MetaReferenceEvalConfig",
             api_dependencies=[
@@ -28,9 +27,6 @@ def available_providers() -> list[ProviderSpec]:
         RemoteProviderSpec(
             api=Api.eval,
             adapter_type="nvidia",
-            pip_packages=[
-                "requests",
-            ],
             provider_type="remote::nvidia",
             module="llama_stack.providers.remote.eval.nvidia",
             config_class="llama_stack.providers.remote.eval.nvidia.NVIDIAEvalConfig",
diff --git a/llama_stack/providers/registry/files.py b/llama_stack/providers/registry/files.py
index 9acabfacd..7c72ef944 100644
--- a/llama_stack/providers/registry/files.py
+++ b/llama_stack/providers/registry/files.py
@@ -14,7 +14,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.files,
             provider_type="inline::localfs",
             # TODO: make this dynamic according to the sql store type
-            pip_packages=sql_store_pip_packages,
             module="llama_stack.providers.inline.files.localfs",
             config_class="llama_stack.providers.inline.files.localfs.config.LocalfsFilesImplConfig",
             description="Local filesystem-based file storage provider for managing files and documents locally.",
diff --git a/llama_stack/providers/registry/inference.py b/llama_stack/providers/registry/inference.py
index 658611698..4eed0ea09 100644
--- a/llama_stack/providers/registry/inference.py
+++ b/llama_stack/providers/registry/inference.py
@@ -31,7 +31,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.inference,
             provider_type="inline::meta-reference",
-            pip_packages=META_REFERENCE_DEPS,
             module="llama_stack.providers.inline.inference.meta_reference",
             config_class="llama_stack.providers.inline.inference.meta_reference.MetaReferenceInferenceConfig",
             description="Meta's reference implementation of inference with support for various model formats and optimization techniques.",
@@ -39,11 +38,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.inference,
             provider_type="inline::sentence-transformers",
-            # CrossEncoder depends on torchao.quantization
-            pip_packages=[
-                "torch torchvision torchao>=0.12.0 --extra-index-url https://download.pytorch.org/whl/cpu",
-                "sentence-transformers --no-deps",
-            ],
             module="llama_stack.providers.inline.inference.sentence_transformers",
             config_class="llama_stack.providers.inline.inference.sentence_transformers.config.SentenceTransformersInferenceConfig",
             description="Sentence Transformers inference provider for text embeddings and similarity search.",
@@ -52,9 +46,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="cerebras",
             provider_type="remote::cerebras",
-            pip_packages=[
-                "cerebras_cloud_sdk",
-            ],
             module="llama_stack.providers.remote.inference.cerebras",
             config_class="llama_stack.providers.remote.inference.cerebras.CerebrasImplConfig",
             description="Cerebras inference provider for running models on Cerebras Cloud platform.",
@@ -63,7 +54,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="ollama",
             provider_type="remote::ollama",
-            pip_packages=["ollama", "aiohttp", "h11>=0.16.0"],
             config_class="llama_stack.providers.remote.inference.ollama.OllamaImplConfig",
             module="llama_stack.providers.remote.inference.ollama",
             description="Ollama inference provider for running local models through the Ollama runtime.",
@@ -72,7 +62,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="vllm",
             provider_type="remote::vllm",
-            pip_packages=[],
             module="llama_stack.providers.remote.inference.vllm",
             config_class="llama_stack.providers.remote.inference.vllm.VLLMInferenceAdapterConfig",
             provider_data_validator="llama_stack.providers.remote.inference.vllm.VLLMProviderDataValidator",
@@ -82,7 +71,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="tgi",
             provider_type="remote::tgi",
-            pip_packages=["huggingface_hub", "aiohttp"],
             module="llama_stack.providers.remote.inference.tgi",
             config_class="llama_stack.providers.remote.inference.tgi.TGIImplConfig",
             description="Text Generation Inference (TGI) provider for HuggingFace model serving.",
@@ -91,7 +79,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="hf::serverless",
             provider_type="remote::hf::serverless",
-            pip_packages=["huggingface_hub", "aiohttp"],
             module="llama_stack.providers.remote.inference.tgi",
             config_class="llama_stack.providers.remote.inference.tgi.InferenceAPIImplConfig",
             description="HuggingFace Inference API serverless provider for on-demand model inference.",
@@ -100,7 +87,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             provider_type="remote::hf::endpoint",
             adapter_type="hf::endpoint",
-            pip_packages=["huggingface_hub", "aiohttp"],
             module="llama_stack.providers.remote.inference.tgi",
             config_class="llama_stack.providers.remote.inference.tgi.InferenceEndpointImplConfig",
             description="HuggingFace Inference Endpoints provider for dedicated model serving.",
@@ -109,9 +95,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="fireworks",
             provider_type="remote::fireworks",
-            pip_packages=[
-                "fireworks-ai<=0.17.16",
-            ],
             module="llama_stack.providers.remote.inference.fireworks",
             config_class="llama_stack.providers.remote.inference.fireworks.FireworksImplConfig",
             provider_data_validator="llama_stack.providers.remote.inference.fireworks.FireworksProviderDataValidator",
@@ -121,9 +104,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="together",
             provider_type="remote::together",
-            pip_packages=[
-                "together",
-            ],
             module="llama_stack.providers.remote.inference.together",
             config_class="llama_stack.providers.remote.inference.together.TogetherImplConfig",
             provider_data_validator="llama_stack.providers.remote.inference.together.TogetherProviderDataValidator",
@@ -133,7 +113,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="bedrock",
             provider_type="remote::bedrock",
-            pip_packages=["boto3"],
             module="llama_stack.providers.remote.inference.bedrock",
             config_class="llama_stack.providers.remote.inference.bedrock.BedrockConfig",
             description="AWS Bedrock inference provider for accessing various AI models through AWS's managed service.",
@@ -142,7 +121,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="databricks",
             provider_type="remote::databricks",
-            pip_packages=[],
             module="llama_stack.providers.remote.inference.databricks",
             config_class="llama_stack.providers.remote.inference.databricks.DatabricksImplConfig",
             description="Databricks inference provider for running models on Databricks' unified analytics platform.",
@@ -151,7 +129,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="nvidia",
             provider_type="remote::nvidia",
-            pip_packages=[],
             module="llama_stack.providers.remote.inference.nvidia",
             config_class="llama_stack.providers.remote.inference.nvidia.NVIDIAConfig",
             description="NVIDIA inference provider for accessing NVIDIA NIM models and AI services.",
@@ -160,7 +137,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="runpod",
             provider_type="remote::runpod",
-            pip_packages=[],
             module="llama_stack.providers.remote.inference.runpod",
             config_class="llama_stack.providers.remote.inference.runpod.RunpodImplConfig",
             description="RunPod inference provider for running models on RunPod's cloud GPU platform.",
@@ -169,7 +145,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="openai",
             provider_type="remote::openai",
-            pip_packages=["litellm"],
             module="llama_stack.providers.remote.inference.openai",
             config_class="llama_stack.providers.remote.inference.openai.OpenAIConfig",
             provider_data_validator="llama_stack.providers.remote.inference.openai.config.OpenAIProviderDataValidator",
@@ -179,7 +154,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="anthropic",
             provider_type="remote::anthropic",
-            pip_packages=["litellm"],
             module="llama_stack.providers.remote.inference.anthropic",
             config_class="llama_stack.providers.remote.inference.anthropic.AnthropicConfig",
             provider_data_validator="llama_stack.providers.remote.inference.anthropic.config.AnthropicProviderDataValidator",
@@ -189,9 +163,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="gemini",
             provider_type="remote::gemini",
-            pip_packages=[
-                "litellm",
-            ],
             module="llama_stack.providers.remote.inference.gemini",
             config_class="llama_stack.providers.remote.inference.gemini.GeminiConfig",
             provider_data_validator="llama_stack.providers.remote.inference.gemini.config.GeminiProviderDataValidator",
@@ -201,10 +172,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="vertexai",
             provider_type="remote::vertexai",
-            pip_packages=[
-                "litellm",
-                "google-cloud-aiplatform",
-            ],
             module="llama_stack.providers.remote.inference.vertexai",
             config_class="llama_stack.providers.remote.inference.vertexai.VertexAIConfig",
             provider_data_validator="llama_stack.providers.remote.inference.vertexai.config.VertexAIProviderDataValidator",
@@ -233,9 +200,6 @@ Available Models:
             api=Api.inference,
             adapter_type="groq",
             provider_type="remote::groq",
-            pip_packages=[
-                "litellm",
-            ],
             module="llama_stack.providers.remote.inference.groq",
             config_class="llama_stack.providers.remote.inference.groq.GroqConfig",
             provider_data_validator="llama_stack.providers.remote.inference.groq.config.GroqProviderDataValidator",
@@ -245,7 +209,6 @@ Available Models:
             api=Api.inference,
             adapter_type="llama-openai-compat",
             provider_type="remote::llama-openai-compat",
-            pip_packages=["litellm"],
             module="llama_stack.providers.remote.inference.llama_openai_compat",
             config_class="llama_stack.providers.remote.inference.llama_openai_compat.config.LlamaCompatConfig",
             provider_data_validator="llama_stack.providers.remote.inference.llama_openai_compat.config.LlamaProviderDataValidator",
@@ -255,9 +218,6 @@ Available Models:
             api=Api.inference,
             adapter_type="sambanova",
             provider_type="remote::sambanova",
-            pip_packages=[
-                "litellm",
-            ],
             module="llama_stack.providers.remote.inference.sambanova",
             config_class="llama_stack.providers.remote.inference.sambanova.SambaNovaImplConfig",
             provider_data_validator="llama_stack.providers.remote.inference.sambanova.config.SambaNovaProviderDataValidator",
@@ -267,7 +227,6 @@ Available Models:
             api=Api.inference,
             adapter_type="passthrough",
             provider_type="remote::passthrough",
-            pip_packages=[],
             module="llama_stack.providers.remote.inference.passthrough",
             config_class="llama_stack.providers.remote.inference.passthrough.PassthroughImplConfig",
             provider_data_validator="llama_stack.providers.remote.inference.passthrough.PassthroughProviderDataValidator",
@@ -277,7 +236,6 @@ Available Models:
             api=Api.inference,
             adapter_type="watsonx",
             provider_type="remote::watsonx",
-            pip_packages=["ibm_watsonx_ai"],
             module="llama_stack.providers.remote.inference.watsonx",
             config_class="llama_stack.providers.remote.inference.watsonx.WatsonXConfig",
             provider_data_validator="llama_stack.providers.remote.inference.watsonx.WatsonXProviderDataValidator",
@@ -287,7 +245,6 @@ Available Models:
             api=Api.inference,
             provider_type="remote::azure",
             adapter_type="azure",
-            pip_packages=["litellm"],
             module="llama_stack.providers.remote.inference.azure",
             config_class="llama_stack.providers.remote.inference.azure.AzureConfig",
             provider_data_validator="llama_stack.providers.remote.inference.azure.config.AzureProviderDataValidator",
diff --git a/llama_stack/providers/registry/post_training.py b/llama_stack/providers/registry/post_training.py
index 2092e3b2d..851e6990f 100644
--- a/llama_stack/providers/registry/post_training.py
+++ b/llama_stack/providers/registry/post_training.py
@@ -5,15 +5,12 @@
 # the root directory of this source tree.
 
 
-from typing import cast
-
 from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
 
 # We provide two versions of these providers so that distributions can package the appropriate version of torch.
 # The CPU version is used for distributions that don't have GPU support -- they result in smaller container images.
 torchtune_def = dict(
     api=Api.post_training,
-    pip_packages=["numpy"],
     module="llama_stack.providers.inline.post_training.torchtune",
     config_class="llama_stack.providers.inline.post_training.torchtune.TorchtunePostTrainingConfig",
     api_dependencies=[
@@ -27,28 +24,32 @@ torchtune_def = dict(
 def available_providers() -> list[ProviderSpec]:
     return [
         InlineProviderSpec(
-            **{  # type: ignore
-                **torchtune_def,
-                "provider_type": "inline::torchtune-cpu",
-                "pip_packages": (
-                    cast(list[str], torchtune_def["pip_packages"])
-                    + ["torch torchtune>=0.5.0 torchao>=0.12.0 --extra-index-url https://download.pytorch.org/whl/cpu"]
-                ),
-            },
+            api=Api.post_training,
+            provider_type="inline::torchtune-cpu",
+            module="llama_stack.providers.inline.post_training.torchtune",
+            config_class="llama_stack.providers.inline.post_training.torchtune.TorchtunePostTrainingConfig",
+            api_dependencies=[
+                Api.datasetio,
+                Api.datasets,
+            ],
+            description="TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework (CPU).",
+            package_extras=["cpu"],
         ),
         InlineProviderSpec(
-            **{  # type: ignore
-                **torchtune_def,
-                "provider_type": "inline::torchtune-gpu",
-                "pip_packages": (
-                    cast(list[str], torchtune_def["pip_packages"]) + ["torch torchtune>=0.5.0 torchao>=0.12.0"]
-                ),
-            },
+            api=Api.post_training,
+            provider_type="inline::torchtune-gpu",
+            module="llama_stack.providers.inline.post_training.torchtune",
+            config_class="llama_stack.providers.inline.post_training.torchtune.TorchtunePostTrainingConfig",
+            api_dependencies=[
+                Api.datasetio,
+                Api.datasets,
+            ],
+            description="TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework (GPU).",
+            package_extras=["gpu"],
         ),
         InlineProviderSpec(
             api=Api.post_training,
             provider_type="inline::huggingface-gpu",
-            pip_packages=["trl", "transformers", "peft", "datasets>=4.0.0", "torch"],
             module="llama_stack.providers.inline.post_training.huggingface",
             config_class="llama_stack.providers.inline.post_training.huggingface.HuggingFacePostTrainingConfig",
             api_dependencies=[
@@ -61,7 +62,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.post_training,
             adapter_type="nvidia",
             provider_type="remote::nvidia",
-            pip_packages=["requests", "aiohttp"],
             module="llama_stack.providers.remote.post_training.nvidia",
             config_class="llama_stack.providers.remote.post_training.nvidia.NvidiaPostTrainingConfig",
             description="NVIDIA's post-training provider for fine-tuning models on NVIDIA's platform.",
diff --git a/llama_stack/providers/registry/safety.py b/llama_stack/providers/registry/safety.py
index b30074398..0df59e167 100644
--- a/llama_stack/providers/registry/safety.py
+++ b/llama_stack/providers/registry/safety.py
@@ -18,10 +18,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.safety,
             provider_type="inline::prompt-guard",
-            pip_packages=[
-                "transformers[accelerate]",
-                "torch --index-url https://download.pytorch.org/whl/cpu",
-            ],
             module="llama_stack.providers.inline.safety.prompt_guard",
             config_class="llama_stack.providers.inline.safety.prompt_guard.PromptGuardConfig",
             description="Prompt Guard safety provider for detecting and filtering unsafe prompts and content.",
@@ -29,7 +25,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.safety,
             provider_type="inline::llama-guard",
-            pip_packages=[],
             module="llama_stack.providers.inline.safety.llama_guard",
             config_class="llama_stack.providers.inline.safety.llama_guard.LlamaGuardConfig",
             api_dependencies=[
@@ -40,9 +35,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.safety,
             provider_type="inline::code-scanner",
-            pip_packages=[
-                "codeshield",
-            ],
             module="llama_stack.providers.inline.safety.code_scanner",
             config_class="llama_stack.providers.inline.safety.code_scanner.CodeScannerConfig",
             description="Code Scanner safety provider for detecting security vulnerabilities and unsafe code patterns.",
@@ -51,7 +43,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.safety,
             adapter_type="bedrock",
             provider_type="remote::bedrock",
-            pip_packages=["boto3"],
             module="llama_stack.providers.remote.safety.bedrock",
             config_class="llama_stack.providers.remote.safety.bedrock.BedrockSafetyConfig",
             description="AWS Bedrock safety provider for content moderation using AWS's safety services.",
@@ -60,7 +51,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.safety,
             adapter_type="nvidia",
             provider_type="remote::nvidia",
-            pip_packages=["requests"],
             module="llama_stack.providers.remote.safety.nvidia",
             config_class="llama_stack.providers.remote.safety.nvidia.NVIDIASafetyConfig",
             description="NVIDIA's safety provider for content moderation and safety filtering.",
@@ -69,7 +59,6 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.safety,
             adapter_type="sambanova",
             provider_type="remote::sambanova",
-            pip_packages=["litellm", "requests"],
             module="llama_stack.providers.remote.safety.sambanova",
             config_class="llama_stack.providers.remote.safety.sambanova.SambaNovaSafetyConfig",
             provider_data_validator="llama_stack.providers.remote.safety.sambanova.config.SambaNovaProviderDataValidator",
diff --git a/llama_stack/providers/registry/scoring.py b/llama_stack/providers/registry/scoring.py
index a4ec54ed2..1c8367a0d 100644
--- a/llama_stack/providers/registry/scoring.py
+++ b/llama_stack/providers/registry/scoring.py
@@ -13,7 +13,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.scoring,
             provider_type="inline::basic",
-            pip_packages=["requests"],
             module="llama_stack.providers.inline.scoring.basic",
             config_class="llama_stack.providers.inline.scoring.basic.BasicScoringConfig",
             api_dependencies=[
@@ -25,7 +24,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.scoring,
             provider_type="inline::llm-as-judge",
-            pip_packages=[],
             module="llama_stack.providers.inline.scoring.llm_as_judge",
             config_class="llama_stack.providers.inline.scoring.llm_as_judge.LlmAsJudgeScoringConfig",
             api_dependencies=[
@@ -38,7 +36,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.scoring,
             provider_type="inline::braintrust",
-            pip_packages=["autoevals"],
             module="llama_stack.providers.inline.scoring.braintrust",
             config_class="llama_stack.providers.inline.scoring.braintrust.BraintrustScoringConfig",
             api_dependencies=[
diff --git a/llama_stack/providers/registry/telemetry.py b/llama_stack/providers/registry/telemetry.py
index b50b422c1..7f970d6e4 100644
--- a/llama_stack/providers/registry/telemetry.py
+++ b/llama_stack/providers/registry/telemetry.py
@@ -17,10 +17,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.telemetry,
             provider_type="inline::meta-reference",
-            pip_packages=[
-                "opentelemetry-sdk",
-                "opentelemetry-exporter-otlp-proto-http",
-            ],
             optional_api_dependencies=[Api.datasetio],
             module="llama_stack.providers.inline.telemetry.meta_reference",
             config_class="llama_stack.providers.inline.telemetry.meta_reference.config.TelemetryConfig",
diff --git a/llama_stack/providers/registry/tool_runtime.py b/llama_stack/providers/registry/tool_runtime.py
index ad8c31dfd..80a4bd20a 100644
--- a/llama_stack/providers/registry/tool_runtime.py
+++ b/llama_stack/providers/registry/tool_runtime.py
@@ -18,17 +18,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.tool_runtime,
             provider_type="inline::rag-runtime",
-            pip_packages=[
-                "chardet",
-                "pypdf",
-                "tqdm",
-                "numpy",
-                "scikit-learn",
-                "scipy",
-                "nltk",
-                "sentencepiece",
-                "transformers",
-            ],
             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.files],
@@ -40,7 +29,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::brave-search",
             module="llama_stack.providers.remote.tool_runtime.brave_search",
             config_class="llama_stack.providers.remote.tool_runtime.brave_search.config.BraveSearchToolConfig",
-            pip_packages=["requests"],
             provider_data_validator="llama_stack.providers.remote.tool_runtime.brave_search.BraveSearchToolProviderDataValidator",
             description="Brave Search tool for web search capabilities with privacy-focused results.",
         ),
@@ -50,7 +38,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::bing-search",
             module="llama_stack.providers.remote.tool_runtime.bing_search",
             config_class="llama_stack.providers.remote.tool_runtime.bing_search.config.BingSearchToolConfig",
-            pip_packages=["requests"],
             provider_data_validator="llama_stack.providers.remote.tool_runtime.bing_search.BingSearchToolProviderDataValidator",
             description="Bing Search tool for web search capabilities using Microsoft's search engine.",
         ),
@@ -60,7 +47,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::tavily-search",
             module="llama_stack.providers.remote.tool_runtime.tavily_search",
             config_class="llama_stack.providers.remote.tool_runtime.tavily_search.config.TavilySearchToolConfig",
-            pip_packages=["requests"],
             provider_data_validator="llama_stack.providers.remote.tool_runtime.tavily_search.TavilySearchToolProviderDataValidator",
             description="Tavily Search tool for AI-optimized web search with structured results.",
         ),
@@ -70,7 +56,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::wolfram-alpha",
             module="llama_stack.providers.remote.tool_runtime.wolfram_alpha",
             config_class="llama_stack.providers.remote.tool_runtime.wolfram_alpha.config.WolframAlphaToolConfig",
-            pip_packages=["requests"],
             provider_data_validator="llama_stack.providers.remote.tool_runtime.wolfram_alpha.WolframAlphaToolProviderDataValidator",
             description="Wolfram Alpha tool for computational knowledge and mathematical calculations.",
         ),
@@ -80,7 +65,6 @@ def available_providers() -> list[ProviderSpec]:
             provider_type="remote::model-context-protocol",
             module="llama_stack.providers.remote.tool_runtime.model_context_protocol",
             config_class="llama_stack.providers.remote.tool_runtime.model_context_protocol.config.MCPProviderConfig",
-            pip_packages=["mcp>=1.8.1"],
             provider_data_validator="llama_stack.providers.remote.tool_runtime.model_context_protocol.config.MCPProviderDataValidator",
             description="Model Context Protocol (MCP) tool for standardized tool calling and context management.",
         ),
diff --git a/llama_stack/providers/registry/vector_io.py b/llama_stack/providers/registry/vector_io.py
index e8237bc62..76e081bb2 100644
--- a/llama_stack/providers/registry/vector_io.py
+++ b/llama_stack/providers/registry/vector_io.py
@@ -18,7 +18,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::meta-reference",
-            pip_packages=["faiss-cpu"],
             module="llama_stack.providers.inline.vector_io.faiss",
             config_class="llama_stack.providers.inline.vector_io.faiss.FaissVectorIOConfig",
             deprecation_warning="Please use the `inline::faiss` provider instead.",
@@ -29,7 +28,6 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::faiss",
-            pip_packages=["faiss-cpu"],
             module="llama_stack.providers.inline.vector_io.faiss",
             config_class="llama_stack.providers.inline.vector_io.faiss.FaissVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -82,7 +80,6 @@ more details about Faiss in general.
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::sqlite-vec",
-            pip_packages=["sqlite-vec"],
             module="llama_stack.providers.inline.vector_io.sqlite_vec",
             config_class="llama_stack.providers.inline.vector_io.sqlite_vec.SQLiteVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -289,7 +286,6 @@ See [sqlite-vec's GitHub repo](https://github.com/asg017/sqlite-vec/tree/main) f
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::sqlite_vec",
-            pip_packages=["sqlite-vec"],
             module="llama_stack.providers.inline.vector_io.sqlite_vec",
             config_class="llama_stack.providers.inline.vector_io.sqlite_vec.SQLiteVectorIOConfig",
             deprecation_warning="Please use the `inline::sqlite-vec` provider (notice the hyphen instead of underscore) instead.",
@@ -303,7 +299,6 @@ Please refer to the sqlite-vec provider documentation.
             api=Api.vector_io,
             adapter_type="chromadb",
             provider_type="remote::chromadb",
-            pip_packages=["chromadb-client"],
             module="llama_stack.providers.remote.vector_io.chroma",
             config_class="llama_stack.providers.remote.vector_io.chroma.ChromaVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -345,7 +340,6 @@ See [Chroma's documentation](https://docs.trychroma.com/docs/overview/introducti
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::chromadb",
-            pip_packages=["chromadb"],
             module="llama_stack.providers.inline.vector_io.chroma",
             config_class="llama_stack.providers.inline.vector_io.chroma.ChromaVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -389,7 +383,6 @@ See [Chroma's documentation](https://docs.trychroma.com/docs/overview/introducti
             api=Api.vector_io,
             adapter_type="pgvector",
             provider_type="remote::pgvector",
-            pip_packages=["psycopg2-binary"],
             module="llama_stack.providers.remote.vector_io.pgvector",
             config_class="llama_stack.providers.remote.vector_io.pgvector.PGVectorVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -500,7 +493,6 @@ See [PGVector's documentation](https://github.com/pgvector/pgvector) for more de
             api=Api.vector_io,
             adapter_type="weaviate",
             provider_type="remote::weaviate",
-            pip_packages=["weaviate-client"],
             module="llama_stack.providers.remote.vector_io.weaviate",
             config_class="llama_stack.providers.remote.vector_io.weaviate.WeaviateVectorIOConfig",
             provider_data_validator="llama_stack.providers.remote.vector_io.weaviate.WeaviateRequestProviderData",
@@ -541,7 +533,6 @@ See [Weaviate's documentation](https://weaviate.io/developers/weaviate) for more
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::qdrant",
-            pip_packages=["qdrant-client"],
             module="llama_stack.providers.inline.vector_io.qdrant",
             config_class="llama_stack.providers.inline.vector_io.qdrant.QdrantVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -594,7 +585,6 @@ See the [Qdrant documentation](https://qdrant.tech/documentation/) for more deta
             api=Api.vector_io,
             adapter_type="qdrant",
             provider_type="remote::qdrant",
-            pip_packages=["qdrant-client"],
             module="llama_stack.providers.remote.vector_io.qdrant",
             config_class="llama_stack.providers.remote.vector_io.qdrant.QdrantVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -607,7 +597,6 @@ Please refer to the inline provider documentation.
             api=Api.vector_io,
             adapter_type="milvus",
             provider_type="remote::milvus",
-            pip_packages=["pymilvus>=2.4.10"],
             module="llama_stack.providers.remote.vector_io.milvus",
             config_class="llama_stack.providers.remote.vector_io.milvus.MilvusVectorIOConfig",
             api_dependencies=[Api.inference],
@@ -813,7 +802,6 @@ For more details on TLS configuration, refer to the [TLS setup guide](https://mi
         InlineProviderSpec(
             api=Api.vector_io,
             provider_type="inline::milvus",
-            pip_packages=["pymilvus[milvus-lite]>=2.4.10"],
             module="llama_stack.providers.inline.vector_io.milvus",
             config_class="llama_stack.providers.inline.vector_io.milvus.MilvusVectorIOConfig",
             api_dependencies=[Api.inference],
diff --git a/llama_stack/providers/remote/datasetio/huggingface/pyproject.toml b/llama_stack/providers/remote/datasetio/huggingface/pyproject.toml
new file mode 100644
index 000000000..b525f3b97
--- /dev/null
+++ b/llama_stack/providers/remote/datasetio/huggingface/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-datasetio-huggingface"
+version = "0.1.0"
+description = "HuggingFace datasets provider for accessing and managing datasets from the HuggingFace Hub"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "datasets",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/datasetio/nvidia/pyproject.toml b/llama_stack/providers/remote/datasetio/nvidia/pyproject.toml
new file mode 100644
index 000000000..6dc98aec4
--- /dev/null
+++ b/llama_stack/providers/remote/datasetio/nvidia/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-datasetio-nvidia"
+version = "0.1.0"
+description = "NVIDIA's dataset I/O provider for accessing datasets from NVIDIA's data platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "datasets",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/eval/nvidia/pyproject.toml b/llama_stack/providers/remote/eval/nvidia/pyproject.toml
new file mode 100644
index 000000000..969c5930e
--- /dev/null
+++ b/llama_stack/providers/remote/eval/nvidia/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-eval-nvidia"
+version = "0.1.0"
+description = "NVIDIA's evaluation provider for running evaluation tasks on NVIDIA's platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/files/s3/pyproject.toml b/llama_stack/providers/remote/files/s3/pyproject.toml
new file mode 100644
index 000000000..fa92a62c7
--- /dev/null
+++ b/llama_stack/providers/remote/files/s3/pyproject.toml
@@ -0,0 +1,23 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-files-s3"
+version = "0.1.0"
+description = "Local filesystem-based file storage provider for managing files and documents locally"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "boto3",
+    "sqlalchemy[asyncio]",
+    "aiosqlite",
+     "asyncpg",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/anthropic/pyproject.toml b/llama_stack/providers/remote/inference/anthropic/pyproject.toml
new file mode 100644
index 000000000..dbceed308
--- /dev/null
+++ b/llama_stack/providers/remote/inference/anthropic/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-anthropic"
+version = "0.1.0"
+description = "Anthropic inference provider for accessing Claude models and Anthropic's AI services"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/azure/pyproject.toml b/llama_stack/providers/remote/inference/azure/pyproject.toml
new file mode 100644
index 000000000..6f9f66246
--- /dev/null
+++ b/llama_stack/providers/remote/inference/azure/pyproject.toml
@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-azure"
+version = "0.1.0"
+description = "Azure OpenAI inference provider for accessing GPT models and other Azure services. Provider documentation https://learn.microsoft.com/en-us/azure/ai-foundry/openai/overview"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+]
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/bedrock/pyproject.toml b/llama_stack/providers/remote/inference/bedrock/pyproject.toml
new file mode 100644
index 000000000..341c4ffc7
--- /dev/null
+++ b/llama_stack/providers/remote/inference/bedrock/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-bedrock"
+version = "0.1.0"
+description = "AWS Bedrock inference provider for accessing various AI models through AWS's managed service"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "boto3",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/cerebras/pyproject.toml b/llama_stack/providers/remote/inference/cerebras/pyproject.toml
new file mode 100644
index 000000000..3ad34dbe0
--- /dev/null
+++ b/llama_stack/providers/remote/inference/cerebras/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-cerebras"
+version = "0.1.0"
+description = "Cerebras inference provider for running models on Cerebras Cloud platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "cerebras_cloud_sdk",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/databricks/pyproject.toml b/llama_stack/providers/remote/inference/databricks/pyproject.toml
new file mode 100644
index 000000000..1bb551d65
--- /dev/null
+++ b/llama_stack/providers/remote/inference/databricks/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-databricks"
+version = "0.1.0"
+description = "Databricks inference provider for running models on Databricks' unified analytics platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "openai",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/fireworks/pyproject.toml b/llama_stack/providers/remote/inference/fireworks/pyproject.toml
new file mode 100644
index 000000000..6a75e17cc
--- /dev/null
+++ b/llama_stack/providers/remote/inference/fireworks/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-fireworks"
+version = "0.1.0"
+description = "Fireworks AI inference provider for Llama models and other AI models on the Fireworks platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "fireworks-ai",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/gemini/pyproject.toml b/llama_stack/providers/remote/inference/gemini/pyproject.toml
new file mode 100644
index 000000000..37938260b
--- /dev/null
+++ b/llama_stack/providers/remote/inference/gemini/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-gemini"
+version = "0.1.0"
+description = "Google Gemini inference provider for accessing Gemini models and Google's AI services"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/groq/pyproject.toml b/llama_stack/providers/remote/inference/groq/pyproject.toml
new file mode 100644
index 000000000..590b33e38
--- /dev/null
+++ b/llama_stack/providers/remote/inference/groq/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-groq"
+version = "0.1.0"
+description = "Groq inference provider for ultra-fast inference using Groq's LPU technology"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/llama_openai_compat/pyproject.toml b/llama_stack/providers/remote/inference/llama_openai_compat/pyproject.toml
new file mode 100644
index 000000000..b8f45e7db
--- /dev/null
+++ b/llama_stack/providers/remote/inference/llama_openai_compat/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-llama-openai-compat"
+version = "0.1.0"
+description = "Llama OpenAI-compatible provider for using Llama models with OpenAI API format"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/nvidia/pyproject.toml b/llama_stack/providers/remote/inference/nvidia/pyproject.toml
new file mode 100644
index 000000000..6123a32b5
--- /dev/null
+++ b/llama_stack/providers/remote/inference/nvidia/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-nvidia"
+version = "0.1.0"
+description = "NVIDIA inference provider for accessing NVIDIA NIM models and AI services"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "openai",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/ollama/pyproject.toml b/llama_stack/providers/remote/inference/ollama/pyproject.toml
new file mode 100644
index 000000000..89390a612
--- /dev/null
+++ b/llama_stack/providers/remote/inference/ollama/pyproject.toml
@@ -0,0 +1,23 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-ollama"
+version = "0.1.0"
+description = "Ollama inference provider for running local models through the Ollama runtime"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "ollama",
+    "aiohttp",
+    "h11>=0.16.0",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/openai/pyproject.toml b/llama_stack/providers/remote/inference/openai/pyproject.toml
new file mode 100644
index 000000000..471d02571
--- /dev/null
+++ b/llama_stack/providers/remote/inference/openai/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-openai"
+version = "0.1.0"
+description = "OpenAI inference provider for accessing GPT models and other OpenAI services"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/passthrough/pyproject.toml b/llama_stack/providers/remote/inference/passthrough/pyproject.toml
new file mode 100644
index 000000000..57765d81e
--- /dev/null
+++ b/llama_stack/providers/remote/inference/passthrough/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-passthrough"
+version = "0.1.0"
+description = "Passthrough inference provider for connecting to any external inference service not directly supported"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/runpod/pyproject.toml b/llama_stack/providers/remote/inference/runpod/pyproject.toml
new file mode 100644
index 000000000..433991028
--- /dev/null
+++ b/llama_stack/providers/remote/inference/runpod/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-runpod"
+version = "0.1.0"
+description = "RunPod inference provider for running models on RunPod's cloud GPU platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "openai",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/sambanova/pyproject.toml b/llama_stack/providers/remote/inference/sambanova/pyproject.toml
new file mode 100644
index 000000000..0d8318e7c
--- /dev/null
+++ b/llama_stack/providers/remote/inference/sambanova/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-sambanova"
+version = "0.1.0"
+description = "SambaNova inference provider for running models on SambaNova's dataflow architecture"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/tgi/pyproject.toml b/llama_stack/providers/remote/inference/tgi/pyproject.toml
new file mode 100644
index 000000000..4d63b3518
--- /dev/null
+++ b/llama_stack/providers/remote/inference/tgi/pyproject.toml
@@ -0,0 +1,22 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-tgi"
+version = "0.1.0"
+description = "Text Generation Inference (TGI) provider for HuggingFace model serving"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "huggingface_hub",
+    "aiohttp",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/together/pyproject.toml b/llama_stack/providers/remote/inference/together/pyproject.toml
new file mode 100644
index 000000000..9cb290789
--- /dev/null
+++ b/llama_stack/providers/remote/inference/together/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-together"
+version = "0.1.0"
+description = "Together AI inference provider for open-source models and collaborative AI development"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "together",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/vertexai/pyproject.toml b/llama_stack/providers/remote/inference/vertexai/pyproject.toml
new file mode 100644
index 000000000..1b92abb97
--- /dev/null
+++ b/llama_stack/providers/remote/inference/vertexai/pyproject.toml
@@ -0,0 +1,19 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-vertexai"
+version = "0.1.0"
+description = "Google VertexAI Remote Inference Provider"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+    "google-cloud-aiplatform"
+]
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/vllm/pyproject.toml b/llama_stack/providers/remote/inference/vllm/pyproject.toml
new file mode 100644
index 000000000..7a74daa9f
--- /dev/null
+++ b/llama_stack/providers/remote/inference/vllm/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-vllm"
+version = "0.1.0"
+description = "Remote vLLM inference provider for connecting to vLLM servers"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "openai",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/inference/watsonx/pyproject.toml b/llama_stack/providers/remote/inference/watsonx/pyproject.toml
new file mode 100644
index 000000000..6928566ac
--- /dev/null
+++ b/llama_stack/providers/remote/inference/watsonx/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-inference-watsonx"
+version = "0.1.0"
+description = "IBM WatsonX inference provider for accessing AI models on IBM's WatsonX platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "ibm_watson_machine_learning",
+
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/post_training/nvidia/pyproject.toml b/llama_stack/providers/remote/post_training/nvidia/pyproject.toml
new file mode 100644
index 000000000..e94aeb707
--- /dev/null
+++ b/llama_stack/providers/remote/post_training/nvidia/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-post-training-nvidia"
+version = "0.1.0"
+description = "NVIDIA's post-training provider for fine-tuning models on NVIDIA's platform"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+    "aiohttp",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/safety/bedrock/pyproject.toml b/llama_stack/providers/remote/safety/bedrock/pyproject.toml
new file mode 100644
index 000000000..c998cc3ee
--- /dev/null
+++ b/llama_stack/providers/remote/safety/bedrock/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-bedrock"
+version = "0.1.0"
+description = "AWS Bedrock safety provider for content moderation using AWS's safety services"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "boto3",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/safety/nvidia/pyproject.toml b/llama_stack/providers/remote/safety/nvidia/pyproject.toml
new file mode 100644
index 000000000..668dfc641
--- /dev/null
+++ b/llama_stack/providers/remote/safety/nvidia/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-nvidia"
+version = "0.1.0"
+description = "NVIDIA's safety provider for content moderation and safety filtering"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/safety/sambanova/pyproject.toml b/llama_stack/providers/remote/safety/sambanova/pyproject.toml
new file mode 100644
index 000000000..a1c147093
--- /dev/null
+++ b/llama_stack/providers/remote/safety/sambanova/pyproject.toml
@@ -0,0 +1,21 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-safety-sambanova"
+version = "0.1.0"
+description = "SambaNova's safety provider for content moderation and safety filtering"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "litellm",
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/tool_runtime/bing_search/pyproject.toml b/llama_stack/providers/remote/tool_runtime/bing_search/pyproject.toml
new file mode 100644
index 000000000..b2995778a
--- /dev/null
+++ b/llama_stack/providers/remote/tool_runtime/bing_search/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-bing-search"
+version = "0.1.0"
+description = "Bing Search tool for web search capabilities using Microsoft's search engine"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/tool_runtime/brave_search/pyproject.toml b/llama_stack/providers/remote/tool_runtime/brave_search/pyproject.toml
new file mode 100644
index 000000000..ef00f8777
--- /dev/null
+++ b/llama_stack/providers/remote/tool_runtime/brave_search/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-brave-search"
+version = "0.1.0"
+description = "Brave Search tool for web search capabilities with privacy-focused results"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/tool_runtime/model_context_protocol/pyproject.toml b/llama_stack/providers/remote/tool_runtime/model_context_protocol/pyproject.toml
new file mode 100644
index 000000000..a930942d1
--- /dev/null
+++ b/llama_stack/providers/remote/tool_runtime/model_context_protocol/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-model-context-protocol"
+version = "0.1.0"
+description = "Model Context Protocol (MCP) tool for standardized tool calling and context management"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "mcp>=1.8.1",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/tool_runtime/tavily_search/pyproject.toml b/llama_stack/providers/remote/tool_runtime/tavily_search/pyproject.toml
new file mode 100644
index 000000000..55e169614
--- /dev/null
+++ b/llama_stack/providers/remote/tool_runtime/tavily_search/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-tavily-search"
+version = "0.1.0"
+description = "Tavily Search tool for AI-optimized web search with structured results"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/tool_runtime/wolfram_alpha/pyproject.toml b/llama_stack/providers/remote/tool_runtime/wolfram_alpha/pyproject.toml
new file mode 100644
index 000000000..866d77f1b
--- /dev/null
+++ b/llama_stack/providers/remote/tool_runtime/wolfram_alpha/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-tool-runtime-wolfram-alpha"
+version = "0.1.0"
+description = "Wolfram Alpha tool for computational knowledge and mathematical calculations"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "requests",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/vector_io/chroma/pyproject.toml b/llama_stack/providers/remote/vector_io/chroma/pyproject.toml
new file mode 100644
index 000000000..ba728ecb5
--- /dev/null
+++ b/llama_stack/providers/remote/vector_io/chroma/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-chroma-remote"
+version = "0.1.0"
+description = "Chroma remote vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "chromadb-client",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/vector_io/milvus/pyproject.toml b/llama_stack/providers/remote/vector_io/milvus/pyproject.toml
new file mode 100644
index 000000000..66a208604
--- /dev/null
+++ b/llama_stack/providers/remote/vector_io/milvus/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-milvus-remote"
+version = "0.1.0"
+description = "Milvus remote vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "pymilvus>=2.4.10",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/vector_io/pgvector/pyproject.toml b/llama_stack/providers/remote/vector_io/pgvector/pyproject.toml
new file mode 100644
index 000000000..9d761dc8d
--- /dev/null
+++ b/llama_stack/providers/remote/vector_io/pgvector/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-pgvector"
+version = "0.1.0"
+description = "PGVector remote vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "psycopg2-binary",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/vector_io/qdrant/pyproject.toml b/llama_stack/providers/remote/vector_io/qdrant/pyproject.toml
new file mode 100644
index 000000000..f52b91471
--- /dev/null
+++ b/llama_stack/providers/remote/vector_io/qdrant/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-qdrant-remote"
+version = "0.1.0"
+description = "Qdrant remote vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "qdrant-client",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]
diff --git a/llama_stack/providers/remote/vector_io/weaviate/pyproject.toml b/llama_stack/providers/remote/vector_io/weaviate/pyproject.toml
new file mode 100644
index 000000000..811e99c97
--- /dev/null
+++ b/llama_stack/providers/remote/vector_io/weaviate/pyproject.toml
@@ -0,0 +1,20 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "llama-stack-provider-vector-io-weaviate"
+version = "0.1.0"
+description = "Weaviate remote vector database provider for Llama Stack"
+authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
+requires-python = ">=3.12"
+license = { "text" = "MIT" }
+dependencies = [
+    "weaviate-client",
+]
+
+
+
+[tool.setuptools.packages.find]
+where = ["."]
+include = ["llama_stack*"]