academia-mcp 1.9.2__tar.gz → 1.10.0__tar.gz

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: academia-mcp
-Version: 1.9.2
+Version: 1.10.0
 Summary: MCP server that provides different tools to search for scientific publications
 Author-email: Ilya Gusev <phoenixilya@gmail.com>
 Project-URL: Homepage, https://github.com/IlyaGusev/academia_mcp

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp/server.py

@@ -38,7 +38,8 @@ from academia_mcp.tools.bitflip import (
     score_research_proposals,
 )
 from academia_mcp.tools.review import review_pdf_paper, download_pdf_paper
-from academia_mcp.tools.show_image import show_image
+from academia_mcp.tools.show_image import show_image, describe_image
+from academia_mcp.tools.speech_to_text import speech_to_text
 
 
 def configure_uvicorn_style_logging(level: int = logging.INFO) -> None:
@@ -116,11 +117,17 @@ def run(
         server.add_tool(generate_research_proposals)
         server.add_tool(score_research_proposals)
         server.add_tool(document_qa)
+        server.add_tool(describe_image)
         if settings.WORKSPACE_DIR:
             server.add_tool(review_pdf_paper)
     else:
         logger.warning("No OpenRouter API key is set, LLM-related tools will not be available!")
 
+    if settings.OPENAI_API_KEY:
+        server.add_tool(speech_to_text)
+    else:
+        logger.warning("No OpenAI API key is set, speech_to_text will not be available!")
+
     if port is None:
         if settings.PORT is not None:
             port = int(settings.PORT)

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp/settings.py

@@ -11,6 +11,7 @@ class Settings(BaseSettings):
     TAVILY_API_KEY: Optional[str] = None
     EXA_API_KEY: Optional[str] = None
     BRAVE_API_KEY: Optional[str] = None
+    OPENAI_API_KEY: Optional[str] = None
 
     REVIEW_MODEL_NAME: str = "gpt-5"
     BITFLIP_MODEL_NAME: str = "deepseek/deepseek-chat-v3-0324"
@@ -18,6 +19,7 @@ class Settings(BaseSettings):
     DOCUMENT_QA_MODEL_NAME: str = "deepseek/deepseek-chat-v3-0324"
     DOCUMENT_QA_QUESTION_MAX_LENGTH: int = 10000
     DOCUMENT_QA_DOCUMENT_MAX_LENGTH: int = 200000
+    DESCRIBE_IMAGE_MODEL_NAME: str = "gpt-4.1"
 
     PORT: int = 5056
     WORKSPACE_DIR: Optional[Path] = None

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp/tools/__init__.py

@@ -14,7 +14,8 @@ from .web_search import web_search, tavily_web_search, exa_web_search, brave_web
 from .visit_webpage import visit_webpage
 from .bitflip import extract_bitflip_info, generate_research_proposals, score_research_proposals
 from .review import review_pdf_paper, download_pdf_paper, review_pdf_paper_by_url
-from .show_image import show_image
+from .show_image import show_image, describe_image
+from .speech_to_text import speech_to_text
 
 __all__ = [
     "arxiv_search",
@@ -42,4 +43,6 @@ __all__ = [
     "download_pdf_paper",
     "read_pdf",
     "show_image",
+    "describe_image",
+    "speech_to_text",
 ]

academia_mcp-1.10.0/academia_mcp/tools/show_image.py

@@ -0,0 +1,104 @@
+import base64
+from pathlib import Path
+from io import BytesIO
+from typing import Dict, Optional
+from textwrap import dedent
+
+import httpx
+from PIL import Image
+
+from academia_mcp.files import get_workspace_dir
+from academia_mcp.settings import settings
+from academia_mcp.llm import llm_acall, ChatMessage
+
+
+DESCRIBE_PROMPTS = {
+    "general": "Provide a general description of this image. Focus on the main subjects, colors, and overall scene.",
+    "detailed": dedent(
+        """Analyze this image in detail. Include:
+        1. Main subjects and their relationships
+        2. Colors, lighting, and composition
+        3. Any text or symbols present
+        4. Context or possible meaning
+        5. Notable details or interesting elements"""
+    ),
+    "chess": dedent(
+        """Analyze this chess position and provide a detailed description including:
+        1. List of pieces on the board for both white and black
+        2. Whose turn it is to move
+        3. Basic evaluation of the position
+        4. Any immediate tactical opportunities or threats
+        5. Suggested next moves with brief explanations"""
+    ),
+    "text": "Extract and describe any text present in this image. If there are multiple pieces of text, organize them clearly.",
+}
+
+
+def show_image(path: str) -> Dict[str, str]:
+    """
+    Reads an image from the specified URL or from the current work directory.
+    Always call this function at the end of the code block.
+    For instance:
+    ```python
+    show_image("https://example.com/image.png")
+    ```
+    Do not print it ever, just return as the last expression.
+
+    Returns an dictionary with a single "image" key.
+
+    Args:
+        url: Path to file inside current work directory or web URL
+    """
+    if path.startswith("http"):
+        response = httpx.get(path, timeout=10)
+        response.raise_for_status()
+        image = Image.open(BytesIO(response.content))
+    else:
+        assert settings.WORKSPACE_DIR is not None, "WORKSPACE_DIR is not set"
+        full_path = Path(path)
+        if not full_path.exists():
+            full_path = Path(get_workspace_dir()) / path
+            assert full_path.exists(), f"Image file {path} does not exist"
+        image = Image.open(str(full_path))
+    buffer_io = BytesIO()
+    image.save(buffer_io, format="PNG")
+    img_bytes = buffer_io.getvalue()
+    return {"image_base64": base64.b64encode(img_bytes).decode("utf-8")}
+
+
+async def describe_image(
+    path: str, description_type: str = "general", custom_prompt: Optional[str] = None
+) -> str:
+    """
+    Tool to analyze and describe any image using GPT-4 Vision API.
+
+    Returns a description of the image based on the requested type.
+
+    Args:
+        image_path (str): Path to the image file.
+        description_type (str): Type of description to generate. Options:
+            - "general": General description of the image
+            - "detailed": Detailed analysis of the image
+            - "chess": Analysis of a chess position
+            - "text": Extract and describe text from the image
+            - "custom": Custom description based on user prompt
+    """
+    image_base64 = show_image(path)["image_base64"]
+    assert (
+        description_type in DESCRIBE_PROMPTS or description_type == "custom"
+    ), f"Invalid description type: {description_type}"
+    prompt = DESCRIBE_PROMPTS.get(description_type, custom_prompt)
+    assert prompt and prompt.strip(), "Please provide a non-empty prompt"
+    content = [
+        {"type": "text", "text": prompt},
+        {
+            "type": "image_url",
+            "image_url": {"url": f"data:image/jpeg;base64,{image_base64}"},
+        },
+    ]
+    model_name = settings.DESCRIBE_IMAGE_MODEL_NAME
+    response = await llm_acall(
+        model_name=model_name,
+        messages=[ChatMessage(role="user", content=content)],
+    )
+    return response

academia_mcp-1.10.0/academia_mcp/tools/speech_to_text.py

@@ -0,0 +1,48 @@
+from pathlib import Path
+from io import BytesIO
+
+import httpx
+from openai import AsyncOpenAI
+from academia_mcp.files import get_workspace_dir
+
+from academia_mcp.settings import settings
+
+
+async def speech_to_text(audio_path: str, provider: str = "openai") -> str:
+    """
+    Tool to convert speech to text using OpenAI's Whisper model.
+
+    Returns transcribed text from the audio file.
+
+    Args:
+        audio_path (str): Path to the audio file.
+        provider (str): Provider to use. Currently only "openai" is supported.
+    """
+
+    AVAILABLE_PROVIDERS = ("openai",)
+    assert (
+        provider in AVAILABLE_PROVIDERS
+    ), f"Invalid provider: {provider}. Available providers: {AVAILABLE_PROVIDERS}"
+
+    if audio_path.startswith("http"):
+        response = httpx.get(audio_path, timeout=10)
+        response.raise_for_status()
+        ext = audio_path.split(".")[-1]
+        audio_file = BytesIO(response.content)
+        audio_file.name = f"audio_file.{ext}"
+    else:
+        full_audio_path = Path(audio_path)
+        if not full_audio_path.exists():
+            full_audio_path = Path(get_workspace_dir()) / audio_path
+            assert full_audio_path.exists(), f"Audio file {audio_path} does not exist"
+        audio_file = BytesIO(open(full_audio_path, "rb").read())
+        audio_file.name = audio_path.split("/")[-1]
+
+    assert provider == "openai"
+    client = AsyncOpenAI(api_key=settings.OPENAI_API_KEY)
+    result = await client.audio.transcriptions.create(
+        model="gpt-4o-transcribe",
+        file=audio_file,
+        response_format="text",
+    )
+    return result

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp/tools/visit_webpage.py

@@ -8,7 +8,8 @@ from academia_mcp.utils import get_with_retries, post_with_retries
 from academia_mcp.settings import settings
 
 EXA_CONTENTS_URL = "https://api.exa.ai/contents"
-AVAILABLE_PROVIDERS = ("basic", "exa")
+TAVILY_EXTRACT_URL = "https://api.tavily.com/extract"
+AVAILABLE_PROVIDERS = ("basic", "exa", "tavily")
 
 
 def _exa_visit_webpage(url: str) -> str:
@@ -22,7 +23,17 @@ def _exa_visit_webpage(url: str) -> str:
     return json.dumps(response.json()["results"][0])
 
 
-def visit_webpage(url: str, provider: Optional[str] = "basic") -> str:
+def _tavily_visit_webpage(url: str) -> str:
+    key = settings.TAVILY_API_KEY or ""
+    assert key, "Error: TAVILY_API_KEY is not set and no api_key was provided"
+    payload = {
+        "urls": [url],
+    }
+    response = post_with_retries(TAVILY_EXTRACT_URL, payload=payload, api_key=key)
+    return json.dumps(response.json()["results"][0]["raw_content"])
+
+
+def visit_webpage(url: str, provider: Optional[str] = "tavily") -> str:
     """
     Visit a webpage and return the content.
 
@@ -32,7 +43,7 @@ def visit_webpage(url: str, provider: Optional[str] = "basic") -> str:
 
     Args:
         url: The URL of the webpage to visit.
-        provider: The provider to use. Available providers: "basic" (default) or "exa".
+        provider: The provider to use. Available providers: "tavily" (default), "exa", or "basic".
     """
     assert (
         provider in AVAILABLE_PROVIDERS
@@ -40,6 +51,10 @@ def visit_webpage(url: str, provider: Optional[str] = "basic") -> str:
 
     if provider == "exa" and settings.EXA_API_KEY:
         return _exa_visit_webpage(url)
+    elif provider == "tavily" and settings.TAVILY_API_KEY:
+        return _tavily_visit_webpage(url)
+    else:
+        provider = "basic"
 
     assert provider == "basic"
     response = get_with_retries(url)

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: academia-mcp
-Version: 1.9.2
+Version: 1.10.0
 Summary: MCP server that provides different tools to search for scientific publications
 Author-email: Ilya Gusev <phoenixilya@gmail.com>
 Project-URL: Homepage, https://github.com/IlyaGusev/academia_mcp

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/academia_mcp.egg-info/SOURCES.txt

@@ -30,6 +30,7 @@ academia_mcp/tools/py.typed
 academia_mcp/tools/review.py
 academia_mcp/tools/s2.py
 academia_mcp/tools/show_image.py
+academia_mcp/tools/speech_to_text.py
 academia_mcp/tools/visit_webpage.py
 academia_mcp/tools/web_search.py
 tests/test_anthology_search.py
@@ -43,5 +44,6 @@ tests/test_latex.py
 tests/test_review.py
 tests/test_s2.py
 tests/test_show_image.py
+tests/test_speech_to_text.py
 tests/test_visit_webpage.py
 tests/test_web_search.py

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "academia-mcp"
-version = "1.9.2"
+version = "1.10.0"
 description = "MCP server that provides different tools to search for scientific publications"
 readme = "README.md"
 authors = [

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/tests/test_show_image.py

@@ -3,7 +3,7 @@ from io import BytesIO
 import httpx
 from PIL import Image
 
-from academia_mcp.tools import show_image
+from academia_mcp.tools import show_image, describe_image
 from academia_mcp.files import get_workspace_dir
 
 
@@ -23,3 +23,9 @@ def test_show_image_local(test_image_url: str) -> None:
     assert result is not None
     assert "image_base64" in result
     assert result["image_base64"] is not None
+
+
+async def test_describe_image_base(test_image_url: str) -> None:
+    result = await describe_image(test_image_url)
+    assert result is not None
+    assert "Interrogator" in result

academia_mcp-1.10.0/tests/test_speech_to_text.py

@@ -0,0 +1,21 @@
+import httpx
+
+from academia_mcp.tools import speech_to_text
+from academia_mcp.files import get_workspace_dir
+
+
+async def test_speech_to_text_base(test_audio_url: str) -> None:
+    result = await speech_to_text(test_audio_url)
+    assert result is not None
+    assert "dancing in the masquerade" in str(result).lower()
+
+
+async def test_speech_to_text_local(test_audio_url: str) -> None:
+    response = httpx.get(test_audio_url, timeout=10)
+    response.raise_for_status()
+    ext = test_audio_url.split(".")[-1]
+    with open(get_workspace_dir() / f"audio_file.{ext}", "wb") as fp:
+        fp.write(response.content)
+    result = await speech_to_text(f"audio_file.{ext}")
+    assert result is not None
+    assert "dancing in the masquerade" in str(result).lower()

{academia_mcp-1.9.2 → academia_mcp-1.10.0}/tests/test_visit_webpage.py

@@ -18,3 +18,15 @@ def test_visit_webpage_exa() -> None:
 def test_visit_webpage_pdf() -> None:
     content = visit_webpage("https://arxiv.org/pdf/2409.06820")
     assert "A Benchmark for Role-Playing" in content
+
+
+def test_visit_webpage_nature() -> None:
+    url = "https://www.nature.com/nature/articles?page=51&searchType=journalSearch&sort=PubDate&type=article&year=2020"
+    content = visit_webpage(url)
+    assert "1002" in content
+
+
+def test_visit_webpage_github_issue() -> None:
+    url = "https://github.com/numpy/numpy/issues/10881"
+    content = visit_webpage(url, provider="tavily")
+    assert "on Apr 15, 2018" in str(content)

academia_mcp-1.9.2/academia_mcp/tools/show_image.py

@@ -1,41 +0,0 @@
-import base64
-from pathlib import Path
-from io import BytesIO
-from typing import Dict
-
-import httpx
-from PIL import Image
-
-from academia_mcp.files import get_workspace_dir
-from academia_mcp.settings import settings
-
-
-def show_image(path: str) -> Dict[str, str]:
-    """
-    Reads an image from the specified URL or from the current work directory.
-    Always call this function at the end of the code block.
-    For instance:
-    ```python
-    show_image("https://example.com/image.png")
-    ```
-    Do not print it ever, just return as the last expression.
-
-    Returns an dictionary with a single "image" key.
-    Args:
-        url: Path to file inside current work directory or web URL
-    """
-    if path.startswith("http"):
-        response = httpx.get(path, timeout=10)
-        response.raise_for_status()
-        image = Image.open(BytesIO(response.content))
-    else:
-        assert settings.WORKSPACE_DIR is not None, "WORKSPACE_DIR is not set"
-        full_path = Path(path)
-        if not full_path.exists():
-            full_path = Path(get_workspace_dir()) / path
-            assert full_path.exists(), f"Image file {path} does not exist"
-        image = Image.open(str(full_path))
-    buffer_io = BytesIO()
-    image.save(buffer_io, format="PNG")
-    img_bytes = buffer_io.getvalue()
-    return {"image_base64": base64.b64encode(img_bytes).decode("utf-8")}