FACTRAIL
Public beta · Model Context Protocol

Evidence infrastructure
for AI agents.

FACTRAIL gives agents structured, source-backed real-world evidence with provenance, freshness, conflicts, support levels, coverage, and reusable receipts.

Connect via MCPView GitHub and docs
Why FACTRAIL

Evidence an agent can inspect and reuse.

Agents should not have to choose between blindly trusting model output and rebuilding source verification from scratch. FACTRAIL provides a stable evidence layer between agents and real-world data sources. It returns what was established, where it came from, how current it is, what remains unresolved, and a receipt another agent can inspect later.

Capability discoveryfactrail_capabilities exposes supported inputs, fields, sources, and limitations.
Structured Evidencefactrail_verify and factrail_assess return coverage, freshness, support levels, and conflicts.
Reusable receiptsfactrail_get_receipt retrieves content-addressed results and checks their integrity.
Capabilities

Defined coverage. Explicit limits.

Available today · company_fr

French company verification

Verify structured company and establishment information by SIREN or SIRET using INSEE / Sirene and BODACC. Available fields depend on source records. Current registry data takes precedence for current status and legal name; lower-priority evidence remains visible.

factrail_verify
Beta · import

Trade and import assessment

Assess structured import scenarios with caller inputs, source evidence, classification candidates, and deterministic calculations kept distinct. Coverage may be partial or provisional; VAT references are not authoritative EU VAT integration.

factrail_assess
Trust model

Unknown is better than invented.

Sources stay visibleSource identity, support level, freshness, and conflicts travel with the result.
Quality propagatesCaller input is not external verification. Derived results inherit the quality of their inputs.
Failures are explicitMissing input, unsupported scope, stale data, and source failures have distinct outcomes.

A deterministic calculation is not automatically authoritative. Secondary sources are not silently promoted. Receipts let you check content integrity; they are not signatures or claims of absolute truth.

Sources and coverage

Source authority is part of the result.

French company verification uses INSEE / Sirene and BODACC, with field-specific authority handling. Trade/import assessment is beta. FACTRAIL includes validated official TARIC snapshot ingestion infrastructure, but no genuine European Commission snapshot is currently installed. Access2Markets is secondary evidence; curated tariff and VAT references remain curated or provisional. No authoritative EU VAT integration or complete authoritative landed-cost chain is available.

Evidence Contract

Reusable, integrity-checkable receipts.

EvidenceEnvelope reports status, subject, facts, sources, conflicts, coverage, freshness, and a receipt ID. Receipt IDs are content-addressed; retrieval recomputes the canonical hash and rejects altered content. This protects content integrity, not the truth of a claim or a source's correctness.

Developer integration · Public beta

Start with capability discovery.

Connect an MCP client over Streamable HTTP. Call factrail_capabilities first to inspect the deployed operations and limits, then choose a supported verification or assessment.

https://mcp.factrail.online/mcp

factrail_capabilities · factrail_verify · factrail_assess · factrail_get_receipt

GitHub and developer docs ↗Health endpoint ↗

FACTRAIL does not provide arbitrary natural-language fact checking, general web search, broad global company verification, dedicated sanctions or beneficial ownership resolution, authoritative installed EU TARIC data, or signed/blockchain receipts. The live MCP tool list and capability response define current coverage.