Skip to content

OpenRAG is a comprehensive, single package Retrieval-Augmented Generation platform built on Langflow, Docling, and Opensearch.

Python

Add MCP settings/models tools and models API

Edwin Joseb0940fef4793
diff --git a/sdks/mcp/README.md b/sdks/mcp/README.md
index f185e1b4a..ab95550bd 100644
--- a/sdks/mcp/README.md
+++ b/sdks/mcp/README.md
@@ -25,19 +25,57 @@ An MCP (Model Context Protocol) server that exposes OpenRAG capabilities to AI a
2. An OpenRAG API key (create one in Settings → API Keys)
3. Python 3.10+ with `uv` installed
+## Build and run locally
+
+To use the **latest MCP code** (including settings and models tools), run from source. Do **not** install the MCP package if you want local edits to apply.
+
+### What to build (in order)
+
+| Step | What | Command | Required for |
+|------|------|---------|--------------|
+| 1 | OpenRAG backend | Run your OpenRAG app (e.g. frontend + API) | All tools |
+| 2 | MCP from source | `cd sdks/mcp && uv sync` | All tools; no wheel needed |
+| 3 | (Optional) SDK from repo | `cd sdks/python && uv pip install -e .` | Only for latest chat/search/documents SDK |
+
+**Settings and models tools** (`openrag_get_settings`, `openrag_update_settings`, `openrag_list_models`) use direct HTTP and do **not** require the SDK. Chat, search, and document tools use the SDK (PyPI version is fine unless you need unreleased SDK changes).
+
+### Run the MCP from source
+
+```bash
+cd sdks/mcp
+uv sync
+export OPENRAG_URL="http://localhost:3000"
+export OPENRAG_API_KEY="orag_your_api_key"
+uv run openrag-mcp
+```
+
+### Cursor: use repo path so it runs your code
+
+In `~/.cursor/mcp.json`, set `--directory` to your **actual repo path** so Cursor runs the MCP from source, not an installed package:
+
+```json
+"args": ["run", "--directory", "/Users/edwin.jose/Documents/openrag/sdks/mcp", "openrag-mcp"]
+```
+
+If you previously installed the MCP (`pip install openrag-mcp` or installed a wheel), **uninstall it** so Cursor uses the repo:
+
+```bash
+uv pip uninstall openrag-mcp
+```
+
+Then restart Cursor.
+
## Installation
### Option 1: Run from Source (Recommended for Development)
```bash
-# Clone the OpenRAG repository (if you haven't already)
-git clone https://github.com/your-org/openrag.git
cd openrag/sdks/mcp
-
-# Install dependencies
uv sync
```
+Use `uv run openrag-mcp` from that directory, or point Cursor/Claude `--directory` to this path.
+
### Option 2: Install from GitHub
```bash
diff --git a/sdks/mcp/pyproject.toml b/sdks/mcp/pyproject.toml
index 9f288f2c8..aea5a9579 100644
--- a/sdks/mcp/pyproject.toml
+++ b/sdks/mcp/pyproject.toml
@@ -1,6 +1,6 @@
[project]
name = "openrag-mcp"
-version = "0.1.1"
+version = "0.2.1"
description = "MCP server for OpenRAG - Chat, Search, and Ingest documents"
readme = "README.md"
requires-python = ">=3.10"
diff --git a/sdks/mcp/src/openrag_mcp/tools/__init__.py b/sdks/mcp/src/openrag_mcp/tools/__init__.py
index 0225eac55..fce9de1da 100644
--- a/sdks/mcp/src/openrag_mcp/tools/__init__.py
+++ b/sdks/mcp/src/openrag_mcp/tools/__init__.py
@@ -7,6 +7,7 @@ Import this module to register all tools with the registry.
from openrag_mcp.tools import chat # noqa: F401
from openrag_mcp.tools import search # noqa: F401
from openrag_mcp.tools import documents # noqa: F401
+from openrag_mcp.tools import settings # noqa: F401
# Re-export registry functions for convenience
from openrag_mcp.tools.registry import get_all_tools, get_handler
diff --git a/sdks/mcp/src/openrag_mcp/tools/settings.py b/sdks/mcp/src/openrag_mcp/tools/settings.py
new file mode 100644
index 000000000..d82cc226a
--- /dev/null
+++ b/sdks/mcp/src/openrag_mcp/tools/settings.py
@@ -0,0 +1,266 @@
+"""Settings and models tools for OpenRAG MCP server.
+
+Uses direct HTTP calls so these tools work with any OpenRAG backend,
+without depending on SDK version (models client or settings model fields).
+"""
+
+import logging
+
+from mcp.types import TextContent, Tool
+
+from openrag_mcp.config import get_client
+from openrag_mcp.tools.registry import register_tool
+
+logger = logging.getLogger("openrag-mcp.settings")
+
+VALID_PROVIDERS = ("openai", "anthropic", "ollama", "watsonx")
+
+def _format_http_error(response) -> str:
+ """Extract error message from HTTP response."""
+ try:
+ data = response.json()
+ return data.get("error", response.text) or f"HTTP {response.status_code}"
+ except Exception:
+ return response.text or f"HTTP {response.status_code}"
+
+
+async def _request_get(path_with_api: str, path_without_api: str):
+ """GET request; try /api/v1/... first, then /v1/... on 404 (backend-only)."""
+ http = get_client()
+ response = await http.get(path_with_api)
+ if response.status_code == 404 and path_with_api.startswith("/api"):
+ response = await http.get(path_without_api)
+ return response
+
+
+async def _request_post(path_with_api: str, path_without_api: str, json_body: dict):
+ """POST request; try /api/v1/... first, then /v1/... on 404."""
+ http = get_client()
+ response = await http.post(path_with_api, json=json_body)
+ if response.status_code == 404 and path_with_api.startswith("/api"):
+ response = await http.post(path_without_api, json=json_body)
+ return response
+
+
+# ============================================================================
+# Tool: openrag_get_settings
+# ============================================================================
+
+GET_SETTINGS_TOOL = Tool(
+ name="openrag_get_settings",
+ description=(
+ "Get the current OpenRAG configuration. Returns LLM provider and model, "
+ "embedding provider and model, chunk settings, document processing options "
+ "(table structure, OCR, picture descriptions), and system prompt."
+ ),
+ inputSchema={
+ "type": "object",
+ "properties": {},
+ },
+)
+
+
+async def handle_get_settings(arguments: dict) -> list[TextContent]:
+ """Handle openrag_get_settings tool calls."""
+ try:
+ response = await _request_get("/api/v1/settings", "/v1/settings")
+ if response.status_code != 200:
+ return [TextContent(type="text", text=_format_http_error(response))]
+
+ data = response.json()
+ agent = data.get("agent") or {}
+ knowledge = data.get("knowledge") or {}
+
+ parts = ["**Current OpenRAG settings**\n"]
+ parts.append("\n**Agent (LLM)**")
+ parts.append(f"\n- Provider: {agent.get('llm_provider') or '—'}")
+ parts.append(f"\n- Model: {agent.get('llm_model') or '—'}")
+ system_prompt = agent.get("system_prompt")
+ if system_prompt:
+ prompt_preview = (
+ system_prompt[:200] + "..." if len(system_prompt) > 200 else system_prompt
+ )
+ parts.append(f"\n- System prompt: {prompt_preview}")
+ parts.append("\n**Knowledge (embeddings & ingestion)**")
+ parts.append(f"\n- Embedding provider: {knowledge.get('embedding_provider') or '—'}")
+ parts.append(f"\n- Embedding model: {knowledge.get('embedding_model') or '—'}")
+ parts.append(f"\n- Chunk size: {knowledge.get('chunk_size', '—')}")
+ parts.append(f"\n- Chunk overlap: {knowledge.get('chunk_overlap', '—')}")
+ if "table_structure" in knowledge:
+ parts.append(f"\n- Table structure: {knowledge['table_structure']}")
+ if "ocr" in knowledge:
+ parts.append(f"\n- OCR: {knowledge['ocr']}")
+ if "picture_descriptions" in knowledge:
+ parts.append(f"\n- Picture descriptions: {knowledge['picture_descriptions']}")
+
+ return [TextContent(type="text", text="".join(parts))]
+
+ except Exception as e:
+ logger.error("Get settings error: %s", e)
+ return [TextContent(type="text", text=f"Error getting settings: {str(e)}")]
+
+
+# ============================================================================
+# Tool: openrag_update_settings
+# ============================================================================
+
+UPDATE_SETTINGS_TOOL = Tool(
+ name="openrag_update_settings",
+ description=(
+ "Update OpenRAG configuration. All parameters are optional; only provided "
+ "fields are changed. Use this to set LLM model, embedding model, chunk size/overlap, "
+ "system prompt, and document processing options (table structure, OCR, picture descriptions)."
+ ),
+ inputSchema={
+ "type": "object",
+ "properties": {
+ "llm_provider": {
+ "type": "string",
+ "description": "LLM provider: openai, anthropic, watsonx, or ollama",
+ },
+ "llm_model": {
+ "type": "string",
+ "description": "Language model name (e.g. gpt-4o, claude-sonnet-4)",
+ },
+ "embedding_provider": {
+ "type": "string",
+ "description": "Embedding provider: openai, watsonx, or ollama",
+ },
+ "embedding_model": {
+ "type": "string",
+ "description": "Embedding model name",
+ },
+ "chunk_size": {
+ "type": "integer",
+ "description": "Chunk size for document ingestion",
+ },
+ "chunk_overlap": {
+ "type": "integer",
+ "description": "Chunk overlap for document ingestion",
+ },
+ "system_prompt": {
+ "type": "string",
+ "description": "System prompt for the agent",
+ },
+ "table_structure": {
+ "type": "boolean",
+ "description": "Enable table structure extraction in documents",
+ },
+ "ocr": {
+ "type": "boolean",
+ "description": "Enable OCR for images in documents",
+ },
+ "picture_descriptions": {
+ "type": "boolean",
+ "description": "Enable picture/image descriptions",
+ },
+ },
+ },
+)
+
+
+async def handle_update_settings(arguments: dict) -> list[TextContent]:
+ """Handle openrag_update_settings tool calls."""
+ options = {k: v for k, v in arguments.items() if v is not None}
+ if not options:
+ return [TextContent(type="text", text="No settings to update. Provide at least one option.")]
+
+ try:
+ response = await _request_post("/api/v1/settings", "/v1/settings", options)
+ if response.status_code != 200:
+ return [TextContent(type="text", text=_format_http_error(response))]
+ data = response.json()
+ message = data.get("message", "Configuration updated successfully")
+ return [TextContent(type="text", text=message)]
+ except Exception as e:
+ logger.error("Update settings error: %s", e)
+ return [TextContent(type="text", text=f"Error updating settings: {str(e)}")]
+
+
+# ============================================================================
+# Tool: openrag_list_models
+# ============================================================================
+
+LIST_MODELS_TOOL = Tool(
+ name="openrag_list_models",
+ description=(
+ "List available language models and embedding models for a provider. "
+ "Use this before updating settings to see which model values are valid. "
+ "Provider must be configured in OpenRAG (API key or endpoint set in Settings)."
+ ),
+ inputSchema={
+ "type": "object",
+ "properties": {
+ "provider": {
+ "type": "string",
+ "description": "Provider to list models for: openai, anthropic, ollama, or watsonx",
+ "enum": list(VALID_PROVIDERS),
+ },
+ },
+ "required": ["provider"],
+ },
+)
+
+
+async def handle_list_models(arguments: dict) -> list[TextContent]:
+ """Handle openrag_list_models tool calls."""
+ provider = (arguments.get("provider") or "").lower()
+ if not provider:
+ return [TextContent(type="text", text="Error: provider is required")]
+ if provider not in VALID_PROVIDERS:
+ return [
+ TextContent(
+ type="text",
+ text=f"Error: provider must be one of {', '.join(VALID_PROVIDERS)}",
+ )
+ ]
+
+ try:
+ response = await _request_get(f"/api/v1/models/{provider}", f"/v1/models/{provider}")
+ if response.status_code != 200:
+ return [TextContent(type="text", text=_format_http_error(response))]
+
+ data = response.json()
+ language_models = data.get("language_models") or []
+ embedding_models = data.get("embedding_models") or []
+
+ parts = [f"**Available models for {provider}**\n"]
+ parts.append("\n**Language models**")
+ if language_models:
+ for m in language_models:
+ default = " (default)" if m.get("default") else ""
+ parts.append(f"\n- {m.get('value', m.get('label', ''))}{default}")
+ else:
+ parts.append("\n- None")
+ parts.append("\n**Embedding models**")
+ if embedding_models:
+ for m in embedding_models:
+ default = " (default)" if m.get("default") else ""
+ parts.append(f"\n- {m.get('value', m.get('label', ''))}{default}")
+ else:
+ parts.append("\n- None")
+
+ return [TextContent(type="text", text="".join(parts))]
+
+ except AttributeError as e:
+ if "models" in str(e) and "OpenRAGClient" in str(e):
+ msg = (
+ "You're running an old build of openrag-mcp. To fix:\n"
+ "1. Uninstall: uv pip uninstall openrag-mcp\n"
+ "2. In Cursor MCP config (~/.cursor/mcp.json) use: \"args\": [\"run\", \"--directory\", \"/Users/edwin.jose/Documents/openrag/sdks/mcp\", \"openrag-mcp\"]\n"
+ "3. Restart Cursor."
+ )
+ return [TextContent(type="text", text=msg)]
+ raise
+ except Exception as e:
+ logger.error("List models error: %s", e)
+ return [TextContent(type="text", text=f"Error listing models: {str(e)}")]
+
+
+# ============================================================================
+# Register all tools
+# ============================================================================
+
+register_tool(GET_SETTINGS_TOOL, handle_get_settings)
+register_tool(UPDATE_SETTINGS_TOOL, handle_update_settings)
+register_tool(LIST_MODELS_TOOL, handle_list_models)
diff --git a/sdks/mcp/uv.lock b/sdks/mcp/uv.lock
index 3ce3375fa..d44a91198 100644
--- a/sdks/mcp/uv.lock
+++ b/sdks/mcp/uv.lock
@@ -1,5 +1,4 @@
version = 1
-revision = 1
requires-python = ">=3.10"
[[package]]
@@ -130,7 +129,7 @@ name = "click"
version = "8.3.1"
source = { registry = "https://pypi.org/simple" }
dependencies = [
- { name = "colorama", marker = "sys_platform == 'win32'" },
+ { name = "colorama", marker = "platform_system == 'Windows'" },
]
sdist = { url = "https://files.pythonhosted.org/packages/3d/fa/656b739db8587d7b5dfa22e22ed02566950fbfbcdc20311993483657a5c0/click-8.3.1.tar.gz", hash = "sha256:12ff4785d337a1bb490bb7e9c2b1ee5da3112e94a8622f26a6c77f5d2fc6842a", size = 295065 }
wheels = [
@@ -332,7 +331,7 @@ wheels = [
[[package]]
name = "openrag-mcp"
-version = "0.1.0"
+version = "0.2.1"
source = { editable = "." }
dependencies = [
{ name = "httpx" },
@@ -476,6 +475,14 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/5c/96/5fb7d8c3c17bc8c62fdb031c47d77a1af698f1d7a406b0f79aaa1338f9ad/pydantic_core-2.41.5-cp314-cp314t-win32.whl", hash = "sha256:b4ececa40ac28afa90871c2cc2b9ffd2ff0bf749380fbdf57d165fd23da353aa", size = 1988906 },
{ url = "https://files.pythonhosted.org/packages/22/ed/182129d83032702912c2e2d8bbe33c036f342cc735737064668585dac28f/pydantic_core-2.41.5-cp314-cp314t-win_amd64.whl", hash = "sha256:80aa89cad80b32a912a65332f64a4450ed00966111b6615ca6816153d3585a8c", size = 1981607 },
{ url = "https://files.pythonhosted.org/packages/9f/ed/068e41660b832bb0b1aa5b58011dea2a3fe0ba7861ff38c4d4904c1c1a99/pydantic_core-2.41.5-cp314-cp314t-win_arm64.whl", hash = "sha256:35b44f37a3199f771c3eaa53051bc8a70cd7b54f333531c59e29fd4db5d15008", size = 1974769 },
+ { url = "https://files.pythonhosted.org/packages/11/72/90fda5ee3b97e51c494938a4a44c3a35a9c96c19bba12372fb9c634d6f57/pydantic_core-2.41.5-graalpy311-graalpy242_311_native-macosx_10_12_x86_64.whl", hash = "sha256:b96d5f26b05d03cc60f11a7761a5ded1741da411e7fe0909e27a5e6a0cb7b034", size = 2115441 },
+ { url = "https://files.pythonhosted.org/packages/1f/53/8942f884fa33f50794f119012dc6a1a02ac43a56407adaac20463df8e98f/pydantic_core-2.41.5-graalpy311-graalpy242_311_native-macosx_11_0_arm64.whl", hash = "sha256:634e8609e89ceecea15e2d61bc9ac3718caaaa71963717bf3c8f38bfde64242c", size = 1930291 },
+ { url = "https://files.pythonhosted.org/packages/79/c8/ecb9ed9cd942bce09fc888ee960b52654fbdbede4ba6c2d6e0d3b1d8b49c/pydantic_core-2.41.5-graalpy311-graalpy242_311_native-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:93e8740d7503eb008aa2df04d3b9735f845d43ae845e6dcd2be0b55a2da43cd2", size = 1948632 },
+ { url = "https://files.pythonhosted.org/packages/2e/1b/687711069de7efa6af934e74f601e2a4307365e8fdc404703afc453eab26/pydantic_core-2.41.5-graalpy311-graalpy242_311_native-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f15489ba13d61f670dcc96772e733aad1a6f9c429cc27574c6cdaed82d0146ad", size = 2138905 },
+ { url = "https://files.pythonhosted.org/packages/09/32/59b0c7e63e277fa7911c2fc70ccfb45ce4b98991e7ef37110663437005af/pydantic_core-2.41.5-graalpy312-graalpy250_312_native-macosx_10_12_x86_64.whl", hash = "sha256:7da7087d756b19037bc2c06edc6c170eeef3c3bafcb8f532ff17d64dc427adfd", size = 2110495 },
+ { url = "https://files.pythonhosted.org/packages/aa/81/05e400037eaf55ad400bcd318c05bb345b57e708887f07ddb2d20e3f0e98/pydantic_core-2.41.5-graalpy312-graalpy250_312_native-macosx_11_0_arm64.whl", hash = "sha256:aabf5777b5c8ca26f7824cb4a120a740c9588ed58df9b2d196ce92fba42ff8dc", size = 1915388 },
+ { url = "https://files.pythonhosted.org/packages/6e/0d/e3549b2399f71d56476b77dbf3cf8937cec5cd70536bdc0e374a421d0599/pydantic_core-2.41.5-graalpy312-graalpy250_312_native-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c007fe8a43d43b3969e8469004e9845944f1a80e6acd47c150856bb87f230c56", size = 1942879 },
+ { url = "https://files.pythonhosted.org/packages/f7/07/34573da085946b6a313d7c42f82f16e8920bfd730665de2d11c0c37a74b5/pydantic_core-2.41.5-graalpy312-graalpy250_312_native-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:76d0819de158cd855d1cbb8fcafdf6f5cf1eb8e470abe056d5d161106e38062b", size = 2139017 },
{ url = "https://files.pythonhosted.org/packages/e6/b0/1a2aa41e3b5a4ba11420aba2d091b2d17959c8d1519ece3627c371951e73/pydantic_core-2.41.5-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:b5819cd790dbf0c5eb9f82c73c16b39a65dd6dd4d1439dcdea7816ec9adddab8", size = 2103351 },
{ url = "https://files.pythonhosted.org/packages/a4/ee/31b1f0020baaf6d091c87900ae05c6aeae101fa4e188e1613c80e4f1ea31/pydantic_core-2.41.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:5a4e67afbc95fa5c34cf27d9089bca7fcab4e51e57278d710320a70b956d1b9a", size = 1925363 },
{ url = "https://files.pythonhosted.org/packages/e1/89/ab8e86208467e467a80deaca4e434adac37b10a9d134cd2f99b28a01e483/pydantic_core-2.41.5-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ece5c59f0ce7d001e017643d8d24da587ea1f74f6993467d85ae8a5ef9d4f42b", size = 2135615 },
diff --git a/sdks/python/openrag_sdk/client.py b/sdks/python/openrag_sdk/client.py
index 7d0d2e091..2f5cf4dc1 100644
--- a/sdks/python/openrag_sdk/client.py
+++ b/sdks/python/openrag_sdk/client.py
@@ -19,6 +19,31 @@ from .knowledge_filters import KnowledgeFiltersClient
from .search import SearchClient
+class ModelsClient:
+ """Client for listing available models per provider."""
+
+ def __init__(self, client: "OpenRAGClient"):
+ self._client = client
+
+ async def list(self, provider: str):
+ """
+ List available language and embedding models for a provider.
+
+ Args:
+ provider: One of openai, anthropic, ollama, watsonx.
+
+ Returns:
+ ModelsResponse with language_models and embedding_models lists.
+ """
+ from .models import ModelsResponse
+
+ response = await self._client._request(
+ "GET", f"/api/v1/models/{provider}"
+ )
+ data = response.json()
+ return ModelsResponse(**data)
+
+
class SettingsClient:
"""Client for settings operations."""
@@ -132,6 +157,7 @@ class OpenRAGClient:
self.search = SearchClient(self)
self.documents = DocumentsClient(self)
self.settings = SettingsClient(self)
+ self.models = ModelsClient(self)
self.knowledge_filters = KnowledgeFiltersClient(self)
@property
diff --git a/sdks/python/openrag_sdk/models.py b/sdks/python/openrag_sdk/models.py
index d65c98f18..6496332ac 100644
--- a/sdks/python/openrag_sdk/models.py
+++ b/sdks/python/openrag_sdk/models.py
@@ -134,6 +134,7 @@ class AgentSettings(BaseModel):
llm_provider: str | None = None
llm_model: str | None = None
+ system_prompt: str | None = None
class KnowledgeSettings(BaseModel):
@@ -143,6 +144,9 @@ class KnowledgeSettings(BaseModel):
embedding_model: str | None = None
chunk_size: int | None = None
chunk_overlap: int | None = None
+ table_structure: bool | None = None
+ ocr: bool | None = None
+ picture_descriptions: bool | None = None
class SettingsResponse(BaseModel):
@@ -182,6 +186,22 @@ class SettingsUpdateResponse(BaseModel):
message: str
+# Models (list available LLM/embedding models per provider)
+class ModelOption(BaseModel):
+ """A single model option returned by the models list endpoint."""
+
+ value: str
+ label: str
+ default: bool = False
+
+
+class ModelsResponse(BaseModel):
+ """Response from listing models for a provider."""
+
+ language_models: list[ModelOption] = Field(default_factory=list)
+ embedding_models: list[ModelOption] = Field(default_factory=list)
+
+
# Knowledge filter models
class KnowledgeFilterQueryData(BaseModel):
"""Query configuration stored in a knowledge filter."""
diff --git a/src/api/v1/models.py b/src/api/v1/models.py
new file mode 100644
index 000000000..daf583726
--- /dev/null
+++ b/src/api/v1/models.py
@@ -0,0 +1,111 @@
+"""
+Public API v1 Models endpoint.
+
+Lists available LLM and embedding models per provider.
+Uses API key authentication. Uses stored credentials from config.
+"""
+from starlette.requests import Request
+from starlette.responses import JSONResponse
+from config.settings import get_openrag_config
+from utils.logging_config import get_logger
+
+logger = get_logger(__name__)
+
+VALID_PROVIDERS = frozenset({"openai", "anthropic", "ollama", "watsonx"})
+
+
+async def _fetch_models(provider, config, models_service):
+ """Fetch models for the given provider using config credentials. Returns (models_dict, error_response)."""
+ if provider == "openai":
+ api_key = config.providers.openai.api_key
+ if not api_key:
+ return None, JSONResponse(
+ {"error": "OpenAI API key not configured. Set it in Settings."},
+ status_code=400,
+ )
+ models = await models_service.get_openai_models(api_key=api_key)
+ return models, None
+
+ if provider == "anthropic":
+ api_key = config.providers.anthropic.api_key
+ if not api_key:
+ return None, JSONResponse(
+ {"error": "Anthropic API key not configured. Set it in Settings."},
+ status_code=400,
+ )
+ models = await models_service.get_anthropic_models(api_key=api_key)
+ return models, None
+
+ if provider == "ollama":
+ endpoint = config.providers.ollama.endpoint
+ if not endpoint:
+ return None, JSONResponse(
+ {"error": "Ollama endpoint not configured. Set it in Settings."},
+ status_code=400,
+ )
+ models = await models_service.get_ollama_models(endpoint=endpoint)
+ return models, None
+
+ # watsonx
+ api_key = config.providers.watsonx.api_key
+ endpoint = config.providers.watsonx.endpoint
+ project_id = config.providers.watsonx.project_id
+ if not api_key:
+ return None, JSONResponse(
+ {"error": "WatsonX API key not configured. Set it in Settings."},
+ status_code=400,
+ )
+ if not endpoint:
+ return None, JSONResponse(
+ {"error": "WatsonX endpoint not configured. Set it in Settings."},
+ status_code=400,
+ )
+ if not project_id:
+ return None, JSONResponse(
+ {"error": "WatsonX project ID not configured. Set it in Settings."},
+ status_code=400,
+ )
+ models = await models_service.get_ibm_models(
+ endpoint=endpoint, api_key=api_key, project_id=project_id
+ )
+ return models, None
+
+
+async def list_models_endpoint(request: Request, models_service):
+ """
+ List available language and embedding models for a provider.
+
+ GET /v1/models/{provider}
+
+ Path params:
+ provider: One of openai, anthropic, ollama, watsonx
+
+ Response:
+ {
+ "language_models": [{"value": "...", "label": "...", "default": false}],
+ "embedding_models": [{"value": "...", "label": "...", "default": false}]
+ }
+
+ Uses stored credentials from OpenRAG configuration.
+ """
+ provider = (request.path_params.get("provider") or "").lower()
+ if provider not in VALID_PROVIDERS:
+ return JSONResponse(
+ {
+ "error": f"Invalid provider. Must be one of: {', '.join(sorted(VALID_PROVIDERS))}"
+ },
+ status_code=400,
+ )
+
+ try:
+ config = get_openrag_config()
+ models, error_response = await _fetch_models(provider, config, models_service)
+ if error_response is not None:
+ return error_response
+ return JSONResponse(models)
+ except Exception as e:
+ logger.error("Failed to list models for provider %s: %s", provider, str(e))
+ return JSONResponse(
+ {"error": f"Failed to retrieve models: {str(e)}"},
+ status_code=500,
+ )
diff --git a/src/api/v1/settings.py b/src/api/v1/settings.py
index b926eaf16..f94d59a6a 100644
--- a/src/api/v1/settings.py
+++ b/src/api/v1/settings.py
@@ -43,12 +43,16 @@ async def get_settings_endpoint(request: Request):
"agent": {
"llm_provider": config.agent.llm_provider,
"llm_model": config.agent.llm_model,
+ "system_prompt": config.agent.system_prompt,
},
"knowledge": {
"embedding_provider": config.knowledge.embedding_provider,
"embedding_model": config.knowledge.embedding_model,
"chunk_size": config.knowledge.chunk_size,
"chunk_overlap": config.knowledge.chunk_overlap,
+ "table_structure": config.knowledge.table_structure,
+ "ocr": config.knowledge.ocr,
+ "picture_descriptions": config.knowledge.picture_descriptions,
},
}
diff --git a/src/main.py b/src/main.py
index 710b3dab2..fd49af3e1 100644
--- a/src/main.py
+++ b/src/main.py
@@ -58,7 +58,7 @@ from auth_middleware import optional_auth, require_auth
from api_key_middleware import require_api_key
from services.api_key_service import APIKeyService
from api import keys as api_keys
-from api.v1 import chat as v1_chat, search as v1_search, documents as v1_documents, settings as v1_settings, knowledge_filters as v1_knowledge_filters
+from api.v1 import chat as v1_chat, search as v1_search, documents as v1_documents, settings as v1_settings, models as v1_models, knowledge_filters as v1_knowledge_filters
# Configuration and setup
from config.settings import (
@@ -1422,6 +1422,16 @@ async def create_app():
),
methods=["POST"],
),
+ Route(
+ "/v1/models/{provider}",
+ require_api_key(services["api_key_service"])(
+ partial(
+ v1_models.list_models_endpoint,
+ models_service=services["models_service"],
+ )
+ ),
+ methods=["GET"],
+ ),
# Knowledge filters endpoints
Route(
"/v1/knowledge-filters",