# Knowledge Graph Causal Discovery MCP (`ryanclinton/knowledge-graph-causal-discovery-mcp`) Actor Construct causal graphs from multi-domain data, apply do-calculus reasoning, and estimate causal effects via semiparametric methods -- all through a single MCP interface. - **URL**: https://apify.com/ryanclinton/knowledge-graph-causal-discovery-mcp.md - **Developed by:** [ryan clinton](https://apify.com/ryanclinton) (community) - **Categories:** AI, Developer tools - **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, NaN bookmarks - **User rating**: No ratings yet ## Pricing Pay per event + usage This Actor is paid per event and usage. You are charged both the fixed price for specific events and for Apify platform usage. Learn more: https://docs.apify.com/platform/actors/running/actors-in-store#pay-per-event ## What's an Apify Actor? Actors are a software tools running on the Apify platform, for all kinds of web data extraction and automation use cases. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Actors are written with capital "A". ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. The best way to integrate Actors is as follows. In JavaScript/TypeScript projects, use official [JavaScript/TypeScript client](https://docs.apify.com/api/client/js.md): ```bash npm install apify-client ``` In Python projects, use official [Python client library](https://docs.apify.com/api/client/python.md): ```bash pip install apify-client ``` In shell scripts, use [Apify CLI](https://docs.apify.com/cli/docs.md): ````bash # MacOS / Linux curl -fsSL https://apify.com/install-cli.sh | bash # Windows irm https://apify.com/install-cli.ps1 | iex ```bash In AI frameworks, you might use the [Apify MCP server](https://docs.apify.com/platform/integrations/mcp.md). If your project is in a different language, use the [REST API](https://docs.apify.com/api/v2.md). For usage examples, see the [API](#api) section below. For more details, see Apify documentation as [Markdown index](https://docs.apify.com/llms.txt) and [Markdown full-text](https://docs.apify.com/llms-full.txt). # README ## Knowledge Graph Causal Discovery MCP Server **Construct causal graphs from multi-domain data**, apply **do-calculus reasoning**, and estimate **causal effects via semiparametric methods** -- all through a single MCP interface. This server orchestrates **17 Apify actors** across academic, biomedical, regulatory, economic, and safety data sources, then applies rigorous causal inference algorithms including **FCI structure discovery**, **TMLE estimation**, **RotatE knowledge graph embeddings**, and **sheaf cohomology consistency checking**. Every tool returns structured scoring with mathematical foundations. ### What data can you access? | Data Point | Source | Coverage | |------------|--------|----------| | Academic papers and citations | OpenAlex, Semantic Scholar, Crossref | Global research output | | Preprints and open access | ArXiv, CORE | Physics, CS, biology, math | | Biomedical literature | PubMed | 36M+ biomedical citations | | Clinical trials | ClinicalTrials.gov | 450K+ registered trials | | NIH research grants | NIH Reporter | Active and historical NIH funding | | Drug adverse events | OpenFDA | FDA adverse event reports | | Federal regulations | Federal Register | US regulatory actions | | Congressional legislation | Congress.gov | Bills and resolutions | | Government datasets | Data.gov | Federal open data | | Economic indicators | FRED | Federal Reserve economic data | | World development indicators | World Bank | 200+ country indicators | | Product recalls | CPSC | Consumer product safety recalls | | Consumer complaints | CFPB | Financial consumer complaints | | Encyclopedia context | Wikipedia | Background knowledge | ### MCP Tools | Tool | Price | Description | |------|-------|-------------| | `discover_causal_structure` | $0.10 | Discover causal structure via FCI with KCI tests and GES scored by BIC. Returns directed/bidirectional edges, Markov equivalence class, and BIC score. | | `compute_interventional_effects` | $0.10 | Compute P(Y\|do(X)) via do-calculus with back-door/front-door criteria and Balke-Pearl LP bounds for unidentified effects. | | `simulate_counterfactuals` | $0.10 | Simulate counterfactual outcomes via twin network method. Returns probability of necessity (PN) and sufficiency (PS) with monotonicity checks. | | `extract_causal_claims_literature` | $0.10 | Extract causal claims from academic literature via NLP pattern matching and evidence level classification (RCT, observational, case study, review). | | `embed_causal_knowledge_graph` | $0.10 | Embed causal knowledge graph via RotatE complex-valued embeddings. Returns entity embeddings, link prediction metrics (MRR, Hits@10), and clusters. | | `estimate_causal_effect_tmle` | $0.10 | Estimate causal effects via TMLE with Super Learner ensemble. Returns ATE, influence function norms, and cross-validated risk. | | `check_graph_consistency` | $0.10 | Check causal graph consistency via sheaf cohomology. Tests acyclicity, faithfulness, causal sufficiency, instrument validity, and Markov compatibility. | | `attribute_source_contribution` | $0.10 | Attribute data source contributions to causal graph quality via Shapley values. Returns marginal contributions, nucleolus, and core stability. | ### Data Sources - **OpenAlex** -- Open scholarly metadata covering 250M+ works with citation graphs and author disambiguation - **Semantic Scholar** -- AI-powered academic search with citation context and influential citation detection - **Crossref** -- DOI registration agency with 150M+ metadata records and reference linking - **ArXiv** -- Open access preprint server for physics, mathematics, computer science, and quantitative biology - **CORE** -- Aggregator of 300M+ open access research papers from repositories worldwide - **PubMed** -- MEDLINE biomedical literature database with MeSH indexing and structured abstracts - **ClinicalTrials.gov** -- Registry of 450K+ clinical studies with protocol details and outcome data - **NIH Reporter** -- NIH-funded research grants with investigator, institution, and project details - **OpenFDA Drug Events** -- FDA adverse event reporting system (FAERS) for drug safety surveillance - **Federal Register** -- Daily journal of US government regulatory actions and proposed rules - **Congress Bill Tracker** -- Legislative tracking for bills, resolutions, and amendments in Congress - **Data.gov** -- Federal open data platform with 300K+ government datasets - **FRED Economic Data** -- Federal Reserve economic time series including GDP, inflation, employment, and rates - **World Bank Data** -- Development indicators across 200+ countries covering health, economics, and infrastructure - **CPSC Recall Search** -- Consumer Product Safety Commission recall notices and safety alerts - **CFPB Complaints** -- Consumer Financial Protection Bureau complaint database with company responses - **Wikipedia** -- Encyclopedia articles providing background context and entity descriptions ### How the scoring works Each tool applies a specific **causal inference algorithm** to the knowledge graph built from multi-source data. **Structure discovery** uses FCI (Fast Causal Inference) tolerant of latent confounders, combined with GES (Greedy Equivalence Search) scored by BIC. Additive noise model orientation uses HSIC independence between residuals and cause. **Interventional effects** apply Pearl's three do-calculus rules and the ID algorithm. Back-door and front-door adjustment criteria identify estimable effects. Balke-Pearl LP bounds constrain unidentified effects via linear programming. **Counterfactual simulation** uses the twin network method with factual world (X=x, Y=y) and counterfactual world (X=x') sharing exogenous variables. Tian-Pearl monotonicity bounds are validated. **TMLE estimation** follows the targeted maximum likelihood pipeline: Super Learner initial estimate, propensity score with positivity truncation, clever covariate, targeting step, and influence-function-based confidence intervals. | Score Dimension | Range | Interpretation | |----------------|-------|----------------| | BIC Score | Lower is better | Model fit quality for causal structure | | Branching Ratio | 0.0 - 1.0 | Stability of causal graph (< 1 = stable) | | ATE (Average Treatment Effect) | Varies | Estimated causal effect magnitude | | MRR (Mean Reciprocal Rank) | 0.0 - 1.0 | Knowledge graph embedding quality | | Shapley Value | 0.0 - 1.0 | Data source contribution proportion | ### How to connect this MCP server #### Claude Desktop Add to your `claude_desktop_config.json`: ```json { "mcpServers": { "knowledge-graph-causal-discovery": { "url": "https://knowledge-graph-causal-discovery-mcp.apify.actor/mcp" } } } ```` #### Programmatic (cURL) ```bash curl -X POST https://knowledge-graph-causal-discovery-mcp.apify.actor/mcp \ -H "Content-Type: application/json" \ -H "Authorization: Bearer YOUR_APIFY_TOKEN" \ -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"discover_causal_structure","arguments":{"query":"smoking lung cancer","sources":["academic","biomedical"]}},"id":1}' ``` #### Other MCP clients This server works with any MCP-compatible client including **Cursor**, **Windsurf**, **Cline**, and custom MCP integrations. Point your client to the endpoint `https://knowledge-graph-causal-discovery-mcp.apify.actor/mcp`. ### Use cases for causal discovery intelligence #### Drug safety signal detection Discover causal links between pharmaceutical compounds and adverse events by combining PubMed literature, clinical trial data, and FDA adverse event reports into a unified causal graph. #### Policy impact assessment Estimate the causal effect of regulatory interventions on economic outcomes by combining Federal Register rules, FRED economic indicators, and World Bank development data. #### Systematic review automation Extract and classify causal claims across thousands of academic papers, detecting conflicting evidence and grading claim strength from RCTs down to case studies. #### Knowledge graph completion Use RotatE embeddings to predict missing causal links in biomedical knowledge graphs, identifying undiscovered drug-disease or gene-pathway relationships. #### Data source prioritization Apply Shapley value attribution to determine which data sources contribute most to causal inference quality, guiding data acquisition and budget allocation decisions. #### Treatment effect estimation Use TMLE with Super Learner ensembles to estimate average treatment effects from observational data, producing doubly-robust confidence intervals. ### How much does it cost? This MCP uses **pay-per-event** pricing. You are only charged when a tool is called. Each tool call costs **$0.10**. The Apify Free plan includes **$5 of monthly platform credits**, enough for approximately 50 causal discovery queries per month. | Usage Example | Estimated Cost | |--------------|----------------| | Single causal structure discovery | $0.10 | | Full pipeline (structure + intervention + counterfactual) | $0.30 | | Literature review with claim extraction | $0.10 | | Source attribution analysis across all 5 domains | $0.10 | ### How it works 1. **Tool call received** -- Your MCP client sends a query with optional source category selection (academic, biomedical, regulatory, economic, safety). 2. **Parallel actor execution** -- Up to 17 Apify actors run simultaneously across selected source categories, fetching papers, trials, regulations, economic indicators, and safety data. 3. **Knowledge graph construction** -- Results are merged into a unified causal knowledge graph with nodes (entities, variables, concepts) and edges (causal claims, correlations, references). 4. **Algorithm application** -- The requested causal inference algorithm is applied: FCI/GES for structure, do-calculus for interventions, twin networks for counterfactuals, TMLE for estimation, RotatE for embeddings, or sheaf cohomology for consistency. 5. **Structured response** -- Results are returned as structured JSON with scores, confidence metrics, and supporting evidence. ### FAQ **Q: How fresh is the data?** A: All data is fetched live at query time from each source API. Results reflect the current state of OpenAlex, PubMed, FRED, and all other databases. **Q: Can I select which data sources to use?** A: Yes. Every tool accepts a `sources` parameter with options: `academic`, `biomedical`, `regulatory`, `economic`, and `safety`. You can use any combination. **Q: What is the difference between the structure discovery and interventional tools?** A: Structure discovery finds the causal graph (which variables cause which). Interventional effects compute what happens when you force a variable to a specific value. Use structure discovery first, then interventional analysis. **Q: Is this a replacement for randomized controlled trials?** A: No. This provides causal inference from observational data and literature synthesis. It identifies causal hypotheses and estimates effects, but observational causal inference has inherent limitations compared to experimental designs. **Q: Is it legal to use this data?** A: All data sources are publicly available APIs and open databases. See [Apify's guide on web scraping legality](https://blog.apify.com/is-web-scraping-legal/). **Q: Can I combine this with other MCP servers?** A: Yes. Use this MCP for causal analysis, then feed findings into other intelligence MCPs for domain-specific risk scoring or monitoring. ### Related MCP servers | MCP Server | Focus | |------------|-------| | [ryanclinton/open-source-supply-chain-risk-mcp](https://apify.com/ryanclinton/open-source-supply-chain-risk-mcp) | OSS dependency risk and vulnerability propagation | | [ryanclinton/market-microstructure-manipulation-mcp](https://apify.com/ryanclinton/market-microstructure-manipulation-mcp) | Market manipulation detection with Granger causality | | [ryanclinton/litigation-intelligence-mcp](https://apify.com/ryanclinton/litigation-intelligence-mcp) | Pre-litigation risk scoring from regulatory signals | ### Integrations This MCP server runs on the **Apify platform** and integrates with the broader Apify ecosystem: - **Apify API** -- Call this MCP programmatically from any language via the Apify API - **Scheduling** -- Set up recurring causal analysis runs on a schedule - **Webhooks** -- Trigger downstream workflows when causal analysis completes - **Integrations** -- Connect to Slack, Zapier, Make, or any webhook-compatible service for automated alerts # Actor input Schema ## Actor input object example ```json {} ``` # API You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup. ## JavaScript example ```javascript import { ApifyClient } from 'apify-client'; // Initialize the ApifyClient with your Apify API token // Replace the '' with your token const client = new ApifyClient({ token: '', }); // Prepare Actor input const input = {}; // Run the Actor and wait for it to finish const run = await client.actor("ryanclinton/knowledge-graph-causal-discovery-mcp").call(input); // Fetch and print Actor results from the run's dataset (if any) console.log('Results from dataset'); console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`); const { items } = await client.dataset(run.defaultDatasetId).listItems(); items.forEach((item) => { console.dir(item); }); // 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs ``` ## Python example ```python from apify_client import ApifyClient # Initialize the ApifyClient with your Apify API token # Replace '' with your token. client = ApifyClient("") # Prepare the Actor input run_input = {} # Run the Actor and wait for it to finish run = client.actor("ryanclinton/knowledge-graph-causal-discovery-mcp").call(run_input=run_input) # Fetch and print Actor results from the run's dataset (if there are any) print("💾 Check your data here: https://console.apify.com/storage/datasets/" + run["defaultDatasetId"]) for item in client.dataset(run["defaultDatasetId"]).iterate_items(): print(item) # 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start ``` ## CLI example ```bash echo '{}' | apify call ryanclinton/knowledge-graph-causal-discovery-mcp --silent --output-dataset ``` ## MCP server setup ```json { "mcpServers": { "apify": { "command": "npx", "args": [ "mcp-remote", "https://mcp.apify.com/?tools=ryanclinton/knowledge-graph-causal-discovery-mcp", "--header", "Authorization: Bearer " ] } } } ``` ## OpenAPI specification ```json { "openapi": "3.0.1", "info": { "title": "Knowledge Graph Causal Discovery MCP", "description": "Construct causal graphs from multi-domain data, apply do-calculus reasoning, and estimate causal effects via semiparametric methods -- all through a single MCP interface.", "version": "1.0", "x-build-id": "YpyEK42K1BJy63DuG" }, "servers": [ { "url": "https://api.apify.com/v2" } ], "paths": { "/acts/ryanclinton~knowledge-graph-causal-discovery-mcp/run-sync-get-dataset-items": { "post": { "operationId": "run-sync-get-dataset-items-ryanclinton-knowledge-graph-causal-discovery-mcp", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for its completion, and returns Actor's dataset items in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } }, "/acts/ryanclinton~knowledge-graph-causal-discovery-mcp/runs": { "post": { "operationId": "runs-sync-ryanclinton-knowledge-graph-causal-discovery-mcp", "x-openai-isConsequential": false, "summary": "Executes an Actor and returns information about the initiated run in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK", "content": { "application/json": { "schema": { "$ref": "#/components/schemas/runsResponseSchema" } } } } } } }, "/acts/ryanclinton~knowledge-graph-causal-discovery-mcp/run-sync": { "post": { "operationId": "run-sync-ryanclinton-knowledge-graph-causal-discovery-mcp", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for completion, and returns the OUTPUT from Key-value store in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } } }, "components": { "schemas": { "inputSchema": { "type": "object", "properties": {} }, "runsResponseSchema": { "type": "object", "properties": { "data": { "type": "object", "properties": { "id": { "type": "string" }, "actId": { "type": "string" }, "userId": { "type": "string" }, "startedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "finishedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "status": { "type": "string", "example": "READY" }, "meta": { "type": "object", "properties": { "origin": { "type": "string", "example": "API" }, "userAgent": { "type": "string" } } }, "stats": { "type": "object", "properties": { "inputBodyLen": { "type": "integer", "example": 2000 }, "rebootCount": { "type": "integer", "example": 0 }, "restartCount": { "type": "integer", "example": 0 }, "resurrectCount": { "type": "integer", "example": 0 }, "computeUnits": { "type": "integer", "example": 0 } } }, "options": { "type": "object", "properties": { "build": { "type": "string", "example": "latest" }, "timeoutSecs": { "type": "integer", "example": 300 }, "memoryMbytes": { "type": "integer", "example": 1024 }, "diskMbytes": { "type": "integer", "example": 2048 } } }, "buildId": { "type": "string" }, "defaultKeyValueStoreId": { "type": "string" }, "defaultDatasetId": { "type": "string" }, "defaultRequestQueueId": { "type": "string" }, "buildNumber": { "type": "string", "example": "1.0.0" }, "containerUrl": { "type": "string" }, "usage": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "integer", "example": 1 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } }, "usageTotalUsd": { "type": "number", "example": 0.00005 }, "usageUsd": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "number", "example": 0.00005 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } } } } } } } } } ```