Refactor paper management system and update server tools; enhance research entry structure and streamline JSON data handling.

Author: madebygps

AIResearchHub/data/research_papers.json

@@ -0,0 +1,58 @@
+[
+  {
+    "id": 1,
+    "topic": "few shot learning",
+    "created": "2025-07-14T14:50:54.749520",
+    "status": "complete",
+    "papers_found": [
+      {
+        "title": "EASY: Ensemble Augmented-Shot Y-shaped Learning: State-Of-The-Art Few-Shot Classification with Simple Ingredients",
+        "authors": "Yassir Bendou, Yuqing Hu, Raphael Lafargue, Giulia Lioi, Bastien Pasdeloup, St\u00e9phane Pateux, Vincent Gripon",
+        "url": "https://hf.co/papers/2201.09699"
+      },
+      {
+        "title": "Few-shot Learning with Multilingual Language Models",
+        "authors": "Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal",
+        "url": "https://hf.co/papers/2112.10668"
+      },
+      {
+        "title": "Enhancing Environmental Robustness in Few-shot Learning via Conditional Representation Learning",
+        "authors": "Qianyu Guo, Jingrong Wu, Tianxing Wu, Haofen Wang, Weifeng Ge, Wenqiang Zhang",
+        "url": "https://hf.co/papers/2502.01183"
+      },
+      {
+        "title": "Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference",
+        "authors": "Shell Xu Hu, Da Li, Jan St\u00fchmer, Minyoung Kim, Timothy M. Hospedales",
+        "url": "https://hf.co/papers/2204.07305"
+      }
+    ],
+    "repositories_found": [
+      {
+        "name": "oscarknagg/few-shot",
+        "url": "https://github.com/oscarknagg/few-shot",
+        "stars": 1259
+      },
+      {
+        "name": "sicara/easy-few-shot-learning",
+        "url": "https://github.com/sicara/easy-few-shot-learning",
+        "stars": 1225
+      },
+      {
+        "name": "RL-VIG/LibFewShot",
+        "url": "https://github.com/RL-VIG/LibFewShot",
+        "stars": 1028
+      },
+      {
+        "name": "huggingface/setfit",
+        "url": "https://github.com/huggingface/setfit",
+        "stars": 2522
+      },
+      {
+        "name": "tata1661/FSL-Mate",
+        "url": "https://github.com/tata1661/FSL-Mate",
+        "stars": 1746
+      }
+    ],
+    "notes": "Comprehensive research completed on few-shot learning. Found key papers covering state-of-the-art methods, environmental robustness, multilingual approaches, and simple pipeline improvements. Identified practical implementations including comprehensive libraries (LibFewShot, FSL-Mate), ready-to-use frameworks (SetFit by HuggingFace), and tutorial-focused repositories (easy-few-shot-learning). The research spans both theoretical advances and practical implementation resources."
+  }
+]

AIResearchHub/paper_manager.py

@@ -16,7 +16,7 @@ def __init__(self):
     def load_papers(self) -> list[dict]:
         """Load papers from JSON file."""
         if self.papers_file.exists():
-            return json.loads(self.papers_file.read_text())
+            return json.loads(self.papers_file.read_text()) 
         return []
 
     def save_papers(self, papers: list[dict]):
@@ -51,10 +51,8 @@ def add_paper_to_research(self, research_id: int, paper_data: dict):
                 entry["papers_found"].append({
                     "title": paper_data.get("title", ""),
                     "authors": paper_data.get("authors", ""),
-                    "url": paper_data.get("url", ""),
-                    "added": datetime.now().isoformat()
+                    "url": paper_data.get("url", "")
                 })
-                entry["status"] = "active"
                 break
 
         self.save_papers(papers)
@@ -68,10 +66,8 @@ def add_repo_to_research(self, research_id: int, repo_data: dict):
                 entry["repositories_found"].append({
                     "name": repo_data.get("name", ""),
                     "url": repo_data.get("url", ""),
-                    "stars": repo_data.get("stars", 0),
-                    "added": datetime.now().isoformat()
+                    "stars": repo_data.get("stars", 0)
                 })
-                entry["status"] = "active"
                 break
 
         self.save_papers(papers)
@@ -80,17 +76,8 @@ def get_research_summary(self) -> dict[str, Any]:
         """Get summary of all research activities."""
         papers = self.load_papers()
 
-        if not papers:
-            return {"total_research": 0, "entries": []}
-        
-        status_counts = {}
-        for entry in papers:
-            status = entry.get("status", "pending")
-            status_counts[status] = status_counts.get(status, 0) + 1
-        
         return {
             "total_research": len(papers),
-            "status_breakdown": status_counts,
             "entries": papers,
             "last_updated": datetime.now().isoformat()
         }
@@ -104,7 +91,6 @@ def update_research_status(self, research_id: int, status: str, notes: str = "")
                 entry["status"] = status
                 if notes:
                     entry["notes"] = notes
-                entry["last_updated"] = datetime.now().isoformat()
                 break
 
         self.save_papers(papers)

AIResearchHub/server.py

@@ -1,6 +1,6 @@
 import json
 
-from mcp.server.fastmcp import FastMCP
+from fastmcp import FastMCP
 from paper_manager import PaperManager
 
 # Initialize MCP server and paper manager
@@ -31,31 +31,20 @@ async def research_topic(topic: str) -> dict:
 @mcp.tool(description="Get GitHub search strategies for finding implementations") 
 async def get_github_searches(topic: str) -> dict:
     """
-    Generate optimized GitHub search commands for finding code implementations
+    Generate GitHub search commands for finding code implementations
     
     Args:
         topic: Research topic to find implementations for
         
     Returns:
         GitHub search strategies and commands
     """
-    # Generate targeted search variations
-    searches = [
-        f"{topic} machine learning",
-        f"{topic} python implementation", 
-        f"{topic} pytorch tensorflow",
-        f"{topic} algorithm code"
-    ]
-    
     return {
         "success": True,
         "topic": topic,
-        "github_searches": searches,
-        "commands": [
-            f"Search repos: {topic} stars:>50",
-            f"Search code: {topic} language:python"
-        ],
-        "instructions": "Use GitHub MCP with these search terms to find implementations"
+        "github_searches": [f"{topic} implementation", f"{topic} python"],
+        "commands": [f"Search repos: {topic} stars:>10"],
+        "instructions": "Use GitHub MCP with these search terms"
     }
 
 @mcp.tool(description="Add a paper to a research entry")
@@ -116,62 +105,25 @@ async def add_repository(research_id: int, name: str, url: str = "", stars: int
         "message": f"Repository '{name}' added to research #{research_id}"
     }
 
-@mcp.tool(description="Search local research database")
-async def search_research(query: str) -> dict:
-    """
-    Search your local research database for matching content
-    
-    Args:
-        query: Search term to match against research entries
-        
-    Returns:
-        Matching research entries
-    """
-    papers = paper_manager.load_papers()
-    matching_entries = []
-    
-    for entry in papers:
-        if (query.lower() in entry["topic"].lower() or
-            any(query.lower() in paper.get("title", "").lower() 
-                for paper in entry.get("papers_found", [])) or
-            any(query.lower() in repo.get("name", "").lower() 
-                for repo in entry.get("repositories_found", []))):
-            matching_entries.append(entry)
-    
-    return {
-        "success": True,
-        "query": query,
-        "matches_found": len(matching_entries),
-        "entries": matching_entries
-    }
-
 @mcp.tool(description="Update research status and add notes")
 async def update_research_status(research_id: int, status: str, notes: str = "") -> dict:
     """
     Update the status of a research entry
     
     Args:
         research_id: ID of the research entry to update
-        status: New status (pending, active, complete, archived)
+        status: New status (pending, active, complete)
         notes: Optional notes about the research progress
         
     Returns:
         Success status and details
     """
-    valid_statuses = ["pending", "active", "complete", "archived"]
-    if status not in valid_statuses:
-        return {
-            "success": False,
-            "error": f"Invalid status. Must be one of: {', '.join(valid_statuses)}"
-        }
-    
     paper_manager.update_research_status(research_id, status, notes)
 
     return {
         "success": True,
         "research_id": research_id,
         "status": status,
-        "notes": notes,
         "message": f"Research #{research_id} status updated to '{status}'"
     }
 
@@ -181,46 +133,18 @@ def research_workflow_prompt(topic: str) -> str:
     return f"""Research Topic: {topic}
 
 WORKFLOW:
-1. Use: research_topic(topic="{topic}")
-   - Creates tracking entry with research ID
-
-2. Search for papers using HuggingFace MCP tools:
-   - Use: mcp_huggingface_paper_search(query="{topic}")
-   - For each important paper, use: add_paper(research_id, title, authors, url)
-
-3. Use: get_github_searches(topic="{topic}")
-   - Get optimized search strategies for code implementations
-
-4. Search for repositories using GitHub MCP tools:
-   - Use: mcp_github_search_repositories(query="...")
-   - For each important repo, use: add_repository(research_id, name, url, stars)
-
-5. Use: search_research(query="{topic}")
-   - Review your complete research collection
-
-6. Use: update_research_status(research_id, "complete", "Research completed with X papers and Y repositories")
-   - Mark research as complete when finished
-
-7. Check research://status for your research dashboard
-
-GOAL: Build comprehensive knowledge linking papers with implementations
-NOTE: Remember to save findings and update status when complete!
-
-AI RESEARCH HUB TOOLS:
-- research_topic() - Start new research
-- add_paper() - Save paper findings
-- add_repository() - Save repo findings  
-- get_github_searches() - Get search strategies
-- search_research() - Search saved research
-- update_research_status() - Update status
-
-STATUS OPTIONS:
-- "pending": Just started, no findings yet
-- "active": Currently gathering papers/repos (auto-set when adding content)
-- "complete": Research finished with comprehensive findings
-- "archived": Older research for reference"""
-
-@mcp.resource("research://status")
+1. research_topic(topic="{topic}") - Create research entry
+2. Search HuggingFace papers: mcp_huggingface_paper_search(query="{topic}")
+3. get_github_searches(topic="{topic}") - Get search strategies  
+4. Search GitHub repos: mcp_github_search_repositories(query="{topic}")
+5. Add findings: add_paper() and add_repository()
+6. update_research_status(research_id, "complete") - Mark done
+7. Check research://status for dashboard
+
+TOOLS: research_topic, add_paper, add_repository, get_github_searches, update_research_status
+GOAL: Link papers with implementations"""
+
+@mcp.resource("status://dashboard")
 def research_status() -> str:
     """Current research status and saved topics"""
     summary = paper_manager.get_research_summary()