Allocations MCP Server

Advanced research project discovery with NSF funding integration and enterprise-grade search capabilities. Provides comprehensive access to active research projects and allocations across the ACCESS-CI ecosystem with boolean search operators, smart similarity matching, cross-platform funding analysis, and institutional research profiling.

Usage Examples

Search & Discovery

"Machine learning projects on ACCESS"
"GPU computing allocations from 2024"
"Projects using 'deep learning' NOT 'computer vision'"
"Quantum computing projects with >100K SUs"
"All projects at Stanford"

Similarity & Analysis

"Projects similar to climate modeling research"
"Similar projects to NSF award #2138259"
"Bioinformatics allocations over 500K SUs"
"Projects by PI Dr. Smith sorted by allocation size"

Tools

search_projects

Comprehensive ACCESS-CI research project search and discovery. Supports free-text search, field/resource filtering, similarity matching, and detailed project information retrieval.

Parameters:

• query (string, optional): Search query supporting operators: "term1 AND term2", "term1 OR term2", "term1 NOT term2", exact phrases with quotes. Omit when using project_id or similar_to
• project_id (number, optional): Get detailed information for a specific project ID. Returns full project details including complete abstract
• field_of_science (string, optional): Filter by field of science (e.g., 'Computer Science', 'Physics')
• resource_name (string, optional): Filter projects using specific computational resources (e.g., 'NCSA Delta GPU', 'Purdue Anvil')
• allocation_type (string, optional): Filter by allocation type (e.g., 'Discover', 'Explore', 'Accelerate')
• date_range (object, optional): Filter by project date range with start_date and end_date in YYYY-MM-DD format
• min_allocation (number, optional): Minimum allocation amount filter
• similar_to (number, optional): Find projects similar to this project ID using semantic matching
• similarity_keywords (string, optional): Find projects similar to these keywords/research terms
• similarity_threshold (number, optional): Minimum similarity score (0.0-1.0, default: 0.3)
• include_same_field (boolean, optional): Prioritize projects in same field for similarity search (default: true)
• sort_by (string, optional): Sort by 'relevance', 'date_desc', 'date_asc', 'allocation_desc', 'allocation_asc', 'pi_name' (default: relevance)
• limit (number, optional): Maximum results (default: 20, max: 100)

Examples:

// Search for projects
search_projects({ query: '"machine learning" AND gpu', limit: 10 })

// Get specific project details
search_projects({ project_id: 12345 })

// List by field of science
search_projects({ field_of_science: "Computer Science", limit: 20 })

// Find projects using specific resource
search_projects({ resource_name: "NCSA Delta GPU", limit: 15 })

// Find similar projects
search_projects({ similar_to: 12345, similarity_threshold: 0.7, limit: 10 })

Returns: JSON format {total: number, items: Project[]} where each Project contains:

• projectId, requestNumber, requestTitle
• pi, piInstitution
• fos (field of science)
• abstract
• allocationType, beginDate, endDate
• resources[] (array of allocation resources with names, units, and amounts)
• relevance_score (when searching with query)
• similarity_score (when searching with similarity)

analyze_funding

Comprehensive NSF funding analysis for ACCESS projects and institutions. Cross-references ACCESS allocations with NSF awards to show funding connections, institutional research profiles, and funded project discovery.

Parameters:

• project_id (number, optional): Analyze funding for specific ACCESS project. Returns NSF awards connected to PI and institution
• institution (string, optional): Generate comprehensive funding profile for institution. Shows ACCESS allocations, NSF awards, top researchers, and trends
• pi_name (string, optional): Find funded projects by principal investigator name. Cross-references ACCESS and NSF data
• has_nsf_funding (boolean, optional): Filter to only ACCESS projects with corresponding NSF funding
• field_of_science (string, optional): Filter funded projects by field of science
• limit (number, optional): Maximum projects/awards to return (default: 20 for institution, 10 for others)

Examples:

// Analyze specific project funding
analyze_funding({ project_id: 12345 })

// Generate institutional funding profile
analyze_funding({ institution: "University of Illinois", limit: 20 })

// Find funded projects by PI
analyze_funding({ pi_name: "John Smith", has_nsf_funding: true })

// Find NSF-funded projects in field
analyze_funding({ field_of_science: "Computer Science", has_nsf_funding: true })

Returns: Funding connections, institutional profiles, or funded project lists with ACCESS and NSF data.

get_allocation_statistics

Get aggregate statistics across ACCESS allocations including top 10 fields of science, top 10 most-requested resources, top 10 institutions, and allocation type breakdown. Use for overview/trend questions like 'What are the most popular research areas?' NOT for finding specific projects (use search_projects instead).

Parameters:

• pages_to_analyze (number, optional): Number of pages to analyze for statistics (default: 5, max: 20)

Examples:

// Get allocation statistics
get_allocation_statistics({ pages_to_analyze: 5 })

// Get comprehensive statistics
get_allocation_statistics({ pages_to_analyze: 10 })

Returns: Aggregate statistics including top fields, resources, institutions, and allocation types.

Resources

• accessci://allocations: ACCESS-CI Research Projects and Allocations data

Advanced Search Syntax

Boolean Operators

• AND: "machine learning AND gpu" - Both terms must be present
• OR: "climate OR weather" - Either term can be present
• NOT: "modeling NOT simulation" - First term present, second absent

Exact Phrases

• Quoted strings: "deep learning" - Exact phrase match
• Complex queries: "neural networks" AND gpu NOT tensorflow

Filters & Sorting

• Date ranges: date_range: {start_date: "2024-01-01", end_date: "2024-12-31"}
• Allocation threshold: min_allocation: 50000
• Sort options: sort_by: "allocation_desc" for largest allocations first

API Integration

This server connects to:

• ACCESS-CI Allocations portal: https://allocations.access-ci.org
• NSF Awards MCP server: HTTP communication for enriched funding data
• Inter-server architecture: Enables complex cross-platform analysis

Important Note: ACCESS Credits are computational resource credits, NOT monetary currency.

License

MIT

Installation

npm install -g @access-mcp/allocations

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "access-allocations": {
      "command": "npx",
      "args": ["@access-mcp/allocations"]
    }
  }
}

NSF Integration

The allocations server integrates with the NSF awards server to provide comprehensive funding analysis by cross-referencing ACCESS projects with NSF awards data.

How It Works

The allocations server uses cross-server communication to fetch NSF awards:

1. Allocations Server queries ACCESS projects from allocations.access-ci.org
2. Allocations Server calls NSF Awards Server via HTTP to cross-reference
3. NSF Awards Server queries the official NSF API at api.nsf.gov
4. Results are combined into institutional funding profiles

Local Development Setup

To enable NSF awards cross-referencing in local development, configure both servers in Claude Desktop:

{
  "mcpServers": {
    "nsf-awards": {
      "command": "npx",
      "args": ["-y", "@access-mcp/nsf-awards"],
      "env": {
        "PORT": "3007"
      }
    },
    "allocations": {
      "command": "npx",
      "args": ["-y", "@access-mcp/allocations"],
      "env": {
        "ACCESS_MCP_SERVICES": "nsf-awards=http://localhost:3007"
      }
    }
  }
}

Note: The allocations server works with or without NSF integration:

• With NSF server: Shows comprehensive funding profile (ACCESS + NSF)
• Without NSF server: Shows ACCESS projects only (gracefully degrades)

Production server configuration will be documented in a future update.