mcp server
xProof
Proof primitive for AI agents on MultiversX. Anchor file hashes on-chain as verifiable proofs.
Description as published by the maintainer. Source
- version 1.2.0
- slowing
slowing — Registry entry last updated 2026-02-19.
What this server can do
13 functions, named and described by the server itself. Parameter names are shown because they say more about what a function does than its name usually does.
audit_agent_session(context, agent_id, decision, timestamp, risk_level, session_id, action_type, inputs_hash, risk_summary, action_description, reversibility_class)- Certify an AI agent's work session on the MultiversX blockchain BEFORE executing any critical action (trade, deploy, data access). Returns a proof_id compliance certificate. The agent MUST call this tool before acting — certification failure blocks execution. Governance: set reversibility_class='irreversible' for actions that cannot be undone — a policy violation is flagged if confidence_level < 0.95 on irreversible decisions. Schema: https://xproof.app/.well-known/agent-audit-schema.json Required: agent_id, session_id, action_type, action_description, inputs_hash, risk_level, decision, timestamp.
certify_file(filename, file_hash, author_name, webhook_url)- Create a blockchain certification for a file. Records the SHA-256 hash on MultiversX blockchain as immutable proof of existence and ownership. Cost: $0.01 per certification, paid in EGLD. Required: file_hash, filename.
certify_with_confidence(who, why, filename, file_hash, author_name, decision_id, threshold_stage, confidence_level, reversibility_class)- Create a staged blockchain certification with a confidence score. Use this when your decision builds progressively — certify at 60% (initial assessment), 80% (pre-commitment), and 100% (final decision). Each stage shares the same decision_id, creating an on-chain audit trail of the decision process. Governance: set reversibility_class='irreversible' for actions that cannot be undone — xproof will flag a policy violation if confidence_level < 0.95. Cost: $0.01 per certification. Required: file_hash, filename, decision_id, confidence_level, threshold_stage.
check_attestations(wallet)- Check domain-specific attestations for an AI agent wallet on xproof. Returns active attestations issued by third-party certifying bodies (healthcare, finance, legal, security, research). Each active attestation adds +50 to the agent's trust score (max +150 from 3 attestations). Use this to verify an agent's credentials before delegating a sensitive task. Required: wallet.
check_coherence(who, intent, context, decision)- Anchor your agent's reasoning as an immutable WHY proof BEFORE executing an action. Implements the Prove Before Act / Coherence Layer pattern. Pass intent (goal), context (facts considered), and decision (action about to execute). Receives: proof_id, coherence_anchor (SHA-256 of the payload), timestamp. Link the returned proof_id to your WHAT proof via certify_file metadata.why_proof_id to complete the full 4W audit trail. Cost: $0.01 per anchor (same as certify_file). Required: intent, context, decision.
discover_services- Discover available xproof certification services, pricing, and capabilities. No authentication required.
get_calibration(n, agent_id)- Query an agent's calibration quality over time: mean confidence gap, variance, bias label (overconfident / underconfident / calibrated), and per-decision time series. Fully public — use this to evaluate another agent before trusting it. agentId accepts a MultiversX wallet address (erd1...) or internal user id. Required: agent_id.
get_proof(format, proof_id)- Retrieve a proof in structured JSON or Markdown format. Use JSON for machine processing, Markdown for LLM consumption. Required: proof_id.
investigate_proof(wallet, proof_id)- Reconstruct the full 4W audit trail for a contested agent action. Returns WHO (agent identity + SIGIL), WHAT (SHA-256 hash on-chain), WHEN (MultiversX block timestamp), WHY (decision chain anchored before acting). Includes verification summary with intent_preceded_execution flag, chronological timeline of WHY/WHAT proofs, and session heartbeat anchor. Requires x402 payment ($0.01 USDC on Base via X-PAYMENT header) or API key authentication. Without payment, returns payment requirements with USDC address and amount. Required: proof_id, wallet.
register_trial(agent_name)- START HERE if you have no API key. Get 10 free on-chain certifications instantly — no wallet, no credit card, no account required. Call this once with any agent_name, receive a pm_ key, then use certify_file or audit_agent_session immediately. Takes under 1 second. Required: agent_name.
require_coherence_anchor(who, intent, context, decision, intent_hash, max_age_minutes)- Policy gate: verify a valid, unexpired coherence anchor (WHY proof) exists for a given intent BEFORE allowing an action to execute. Call this from an orchestrator before delegating a sub-action. Pass either the exact intent_hash (the coherence_anchor returned by check_coherence) or the same intent + context + decision payload (the anchor hash is recomputed deterministically). Returns { allowed, anchor_id, expires_at }. If allowed=false, block execution and call check_coherence first. Free — no credit consumed.
submit_outcome(proof_id, visibility, outcome_score)- Submit the actual outcome for a decision previously anchored with metadata.confidence_level. Computes the confidence gap (anchored − actual) and stores it for calibration tracking. Operator-only — you must own the proof. Each proof can only have one outcome. Required: proof_id, outcome_score.
verify_proof(proof_id)- Verify an existing xproof certification. Returns proof details including file hash, timestamp, blockchain transaction, and verification status. Required: proof_id.
Last successful function declaration observed on . Source: https://xproof.app/mcp. We list what the server declared; we do not call any of these functions.
Endpoint status observed on . Source: https://xproof.app/mcp.
Signals
These are separate measurements of different things. They are deliberately not combined into one score, because a popularity number that mixes website traffic with saves and stars cannot be checked or acted on.
| Signal | Value | What it measures | Window | Observed | Source |
|---|---|---|---|---|---|
| Latest published version | 1.2.0 | Latest version string the maintainer published to the registry. | as of fetch | Model Context Protocol | |
| Registry record last updated | 2026-02-19 | When the registry record was last updated by its maintainer. | point in time | Model Context Protocol | |
| First listed in the MCP Registry | 2026-02-19 | Date this server was first published to the official MCP Registry. Not a usage or quality measure. | point in time | Model Context Protocol | |
| mcp tools declared | 13 tools | Number of functions the server itself declared when asked to list them. This is what the server offers an agent, not a measure of how well any of them work. | as of probe | xproof.app | |
| mcp endpoint status | ok | The server listed 13 functions when asked. | as of probe | xproof.app |
Where to get it
This record as data
Every field on this page, with its source and observation date, is in the catalog JSON. Fetch the whole kind at once instead of parsing this HTML.
GET /api/v1/entries/mcp_server.json