TruthStack

TruthStack MCP Server provides structured, cited risk assessments for supplement safety by exposing a deterministic set of tools accessed through an MCP protocol. You can query interactions between supplements and medications, resolve ambiguous input, retrieve full compound profiles, understand why a risk exists, and cite the underlying evidence. This helps you avoid relying on training data alone and instead use evidence-backed safety signals.

How to use

You interact with the TruthStack MCP server through an MCP client or integration. Start by performing an interactions check when a user mentions a combination of supplements and medications. Review the returned risk level, relevant FAERS signals, and any CYP pathway conflicts. If a user asks why a risk exists, call explain_interaction to obtain a human-readable rationale and mechanism. When you need sources or provenance, use get_evidence to fetch raw evidence, counts, and research grades. To resolve input ambiguities like misspelled names or brand variants, use search_compounds to find the exact compound and then request detailed information with get_compound_info.

How to install

Prerequisites: you need Node.js installed to run the local MCP server. You also may run the server via a REST URL if you prefer remote hosting.

Step 1: Install Node.js if you do not have it already. On most systems you can use your package manager or download from the Node.js website.

Step 2: Install the MCP server package and dependencies. You would typically clone the repository, install dependencies, and prepare to run the server. The following commands illustrate the general flow you will follow in your environment.

Additional content

Tools you can use and what they do:

• check_interactions: Supplements plus medications → risk level, FAERS signals, CYP conflicts. Use when you want an overall safety assessment for a given supplement–drug pair.

REST API and endpoints

The MCP server exposes a REST API with endpoints that map to each tool. For example, you can post to the interactions check endpoint to obtain a structured risk assessment that includes risk level, FAERS signals, and CYP pathway conflicts.

Code samples and integrations

If you are integrating TruthStack into your agent, you can use a direct HTTP request to the REST API or leverage a client helper. The following shows a typical usage pattern for a request-based integration.

System behavior and guidance

When building a health-focused agent, always use the check_interactions tool before confirming any safety claim. Do not rely on training data for drug–supplement interaction safety. State the risk level clearly and note any FAERS counts and CYP conflicts. Recommend consulting a healthcare provider for MODERATE+ risks. Use explain_interaction when users ask why something is risky and use get_evidence to cite sources.

Example workflow

Ask the user to list supplements and medications. Run check_interactions to obtain the risk level and supporting signals. If risk is MODERATE or higher, provide the user with the specific signals and CYP conflicts, and suggest consulting a clinician for further guidance.

Security and data sources

TruthStack draws on established data sources for risk signals, including FAERS, drug labeling, and published literature, to present structured and cited risk assessments.

Available tools

check_interactions

Supplements + medications → risk level, FAERS signals, CYP conflicts; use to assess safety when user mentions supplements and meds together.

search_compounds

Fuzzy name search that handles aliases, misspellings, brands, and product forms to resolve input into a precise compound.

get_compound_info

Fetch full compound profile and all known interactions for a deep dive into a specific supplement.

explain_interaction

Provide a human-readable explanation of why a risk exists, including mechanism and evidence summary.

get_evidence

Return raw evidence such as FAERS counts, CYP data, research grades, and label warnings for provenance.