Davisville Labs
AI agent specialist network

Specialist evidence for work that is expensive to get wrong.

Davisville Labs publishes focused, source-backed systems that AI agents can delegate to when raw retrieval is not enough. Each specialist resolves a bounded problem, preserves provenance, exposes uncertainty, and maintains its own domain rules.

Open machine catalog
Encrypted monitoring envelopePublisher manifest
Specialist network

Different domains. One evidence standard.

Davisville Labs is the publisher and routing layer. The work stays with focused specialists that maintain their own evidence rules, source semantics, benchmarks, payment eligibility, and product identity.

Local government intelligenceProduction

LocalProof

California local-government property and jurisdiction intelligence for AI agents.

Use it when
  • A property must be resolved to the correct jurisdiction and official source set.
  • Released parcel, permit, planning, enforcement, zoning, or development evidence needs reconciliation.
  • Source gaps, jurisdiction holds, and uncertainty must stay explicit before paid work.
Visit LocalProofMCP
Bitcoin researchProduction

California Bitcoin

Maintained source-backed Bitcoin research and verification for AI agents.

Use it when
  • A protocol or history question needs maintained evidence.
  • A claim needs contradiction, temporal, or consensus-versus-policy handling.
  • An agent needs an evidence package or timeline rather than generic search results.
Developer entryMCP
SEC intelligenceProduction

FiledProof

Primary-source SEC filing intelligence and reconciliation for AI agents.

Use it when
  • A reported financial fact must be reconciled across filing contexts.
  • Disclosure changes need exact filing evidence and provenance.
  • An agent would otherwise need to reconstruct filing lineage and guarded verification logic.
Visit FiledProofMCP
Environmental evidenceProduction

FacilityProof

Physical-facility environmental compliance evidence across authoritative systems.

Use it when
  • Correct physical-site identity matters before environmental evidence is trusted.
  • Inspections, violations, enforcement, penalties, freshness, and source gaps must be normalized.
  • A wrong-site match or unsafe absence inference would be costly.
  • Available in California, Texas, and Florida for paid full-compliance evidence.
Visit FacilityProofMCP
Film & TV public rightsProduction

FilmRightsProof

U.S. Copyright Office public-record film and television rights intelligence.

Use it when
  • A film or TV property must be resolved against bounded public copyright records.
  • Registrations, recorded documents, renewals, terminations, cancellations, and parties need reconciliation.
  • The result must expose ambiguity and freshness limits rather than inventing clear title.
Visit FilmRightsProofMCP
Pre-payment safetyProduction

ExitProof

Pre-payment safety and reversibility intelligence for autonomous AI-agent commerce.

Use it when
  • An agent should verify a live 402 challenge, price, network, asset, destination, and recovery signals before paying.
  • Cancellation, refundability, deadlines, penalties, or stated loss matter before a real-world commitment.
  • Exit terms should be preserved or later turned into a bounded action pack.
Purchase preflightMCP
Payment identityProduction

PayeeProof

Global public machine-payment identity and payment-destination consistency for AI agents.

Use it when
  • An agent is about to make an autonomous machine payment and needs the endpoint, claimed service, and payment destination to agree.
  • A public HTTP 402, x402 resource binding, MPP challenge, or discovery surface should be reconciled before paying.
  • Unexpected payment-destination or domain changes should be surfaced without treating missing evidence as guilt.
Check a payeeMCP
\n
Agent authorizationProduction

ScopeProof

Machine-verifiable authorization checks for AI agents immediately before consequential actions.

Use it when
  • An agent needs to prove that an exact HTTP, MCP, payment, file, message, shell, or generic action fits a supplied authority envelope.
  • Explicit permissions, deny rules, limits, timing, destructive-action gates, or configured approval requirements must be checked before execution.
  • The decision should be preserved as a signed receipt that another system can independently verify.
Check an actionMCP
Merchant analytics

AI Agent Analytics

For builders and merchants selling to AI agents, measure credible discovery, paid intent, settlement, successful delivery, and repeat use without mixing synthetic traffic with customer demand.

Open AI Agent AnalyticsAgent interfaceMCP
Shared standard

The domains differ. The operating discipline does not.

Every Davisville specialist makes the hard transformation layer explicit instead of charging for commodity public-data retrieval.

Evidence before confidence

Identity, source authority, provenance, conflicts, freshness, and source gaps are part of the product rather than footnotes.

Abstention is a valid result

Missing or ambiguous evidence should stop unsafe conclusions and, where the contract requires it, stop billing before payment.

Compounding state

Resolved identities, adjudicated edge cases, benchmarks, lineage, and source-policy decisions make repeat work stronger over time.

Machine routing

Route by fit, not guesswork.

The machine-readable catalog describes fit, non-fit, production status, public operations, pricing signals, and direct discovery surfaces for every active agent-facing Davisville product.

GET https://davisvillelabs.com/agents/catalog.json
Open machine catalogReceipt Envelope v1