Every change Wellknown observed on this MCP server, newest first, with what it was before and what it became. Tool-surface changes carry the definition diff. Nothing here is edited after the fact.
Changed the definition of "prepare_repair" and "quote_repair"
⟨40 unchanged words⟩ "rail":{"description":"Choose exactly onerailidofferedfrom the descriptors in quote_repair.paymentRails. Saudi quotes normally offer APPLE_PAY only. ⟨27 unchanged words⟩
⟨140 unchanged words⟩ string"},"paymentRails":{"description":"Paymentrailsrail descriptors offered for this quote;choose exactlypass oneofdescriptortheseidintoprepare_repair.prepare_repair.rail.","items":{"additionalProperties":true,"properties":{"id":{"description":"Rail identifier to pass as prepare_repair.rail.","type":"string"}},"required":["id"],"type":"object"},"type":"array"},"phase": ⟨75 unchanged words⟩
Changed the definition of "prepare_repair" and "request_delegated_purchase"
UseonlyONLY for the normal repair flow after quote_repair has already returned a quoteId and the caller explicitlyacceptschoosesathatquoteexistingfromquote.quote_repair.ThisWithtool does not create a quote: with confirm=true,
Added "diagnose_issue", "prepare_repair", "quote_repair" and 2 more; removed "accept_service_quote", "agent_compatibility_check", "api_auth_rescue" and 22 more (25 tools before, 5 now)
Diagnose Issue — Start here. Diagnose one sanitized MCP, API, webhook, agent or integration failure for free. Returns bounded evidence plus a suggested repair service, then use quote_repair only if a priced repair is wanted. It never creates a quote, checkout, payment or repair job. It may record bounded operational attribution/telemetry, so readOnlyHint is false; it does not modify the caller system or commercial state.
Prepare Repair — Use only after the caller explicitly accepts a quote from quote_repair. With confirm=true, records that acceptance and creates checkout instructions on exactly one rail offered by that quote. APPLE_PAY returns a human handoff URL for buyer authorization; BASE_USDC and TRON_USDT return payer transfer instructions. It never sends funds, charges the caller, verifies payment or directly starts repair execution. After the buyer completes the required payment action, use verify_repair with the returned checkoutId and the same rail.
Quote Repair — Use only after diagnose_issue when the caller wants a priced bounded repair. Creates a new non-binding quote for the selected or automatically matched public service. Repeating the call may create a different quoteId, so it is intentionally non-idempotent. It does not accept the quote, create checkout, send payment or start execution. If the quote is accepted by the caller, pass its quoteId and one offered payment rail to prepare_repair.
Added "naif_sos_collision"; changed the definition of "accept_service_quote", "agent_compatibility_check", "api_auth_rescue" and 20 more (24 tools before, 25 now)
RecordUseacceptanceonlyofafteranrequest_service_quoteexistingreturnsservicea quote the caller explicitly chooses to accept. Records that quote as accepted andreturnreturns its eligible paymentrails.rails;
Added "accept_service_quote" and "verify_checkout_payment"; changed the definition of "request_checkout" (22 tools before, 24 now)
Request checkout instructions for an existing accepted quote under the current commerce policy. Choose BASE_USDC or TRON_USDT from paymentRails; default BASE_USDC. A quote can select only one rail. This request does not send funds, charge ⟨8 unchanged words⟩
⟨7 unchanged words⟩ "minLength":8,"type":"string"},"rail":{"enum":["BASE_USDC","TRON_USDT"],"type":"string"},"source":{"maxLength":80,"type":"string ⟨4 unchanged words⟩
Added "json_validity_check" (21 tools before, 22 now)
JSON Parse / Schema Validity Check — Use this first for invalid JSON, malformed JSON, JSON.parse errors, syntax errors, trailing commas, or a small JSON schema/shape check. Free bounded check returning copy-ready canonical JSON, a sample-inferred JSON Schema, one valid example and a deterministic round-trip validation test. It never executes code or creates a quote, checkout or payment.
Added "get_matching_service", "request_checkout" and "request_service_quote" (18 tools before, 21 now)
Get Matching NAIF Service — Match one sanitized technical problem to a public service with its price, currency, delivery estimate and next tool. Read-only; creates no quote or checkout.
Request Eligible Checkout — Request checkout instructions for an existing quote under the current commerce policy. This request does not send funds, charge the client, create a job or execute work.
Request NAIF Service Quote — Create a non-binding quote and receive an explicit eligible next action. Does not create checkout, send payment, create a job or start execution.
Certificate recorded, valid to 2026-12-25
Authorization not required
Removed "accept_agent_offer", "get_agent_market_state", "get_dynamic_agent_offers" and 1 more; changed the definition of "resolve_technical_error" (22 tools before, 18 now)
Sanitize and match an API, EVM RPC,or webhook error to FixGraph. Returns machine-readable repair ⟨5 unchanged words⟩
Accept Agent Offer — Create a real agent-only USDC quote for an active NAIF Market Maker offer. This does not charge or submit payment; checkout remains a separate explicit step.
Get Agent Market State — Read NAIF Agent Market Maker learning, pricing and guardrail state.
Unknown → Live
First tool surface recorded: 22 tools (server version 1.4.0)
https://naifgravity.com/mcp (mcp_streamable_http) — from mcp_registry, with the record
Showing the latest 13 events. The API returns up to 500 and filters by kind: ?kind=tool_surface_changed
CappedUsedelegatedONLY for an owner/platform-preauthorized autonomous purchase of one of four supported micro-services under an explicit spending cap. Unlike prepare_repair, this tool does not consume an existing quoteId: it creates its own quote, accepts it, and creates checkout within the delegation. Repeating the call may create a new quote/checkout, so it is intentionally non-idempotent. Do NOT use it when quote_repair already produced a quoteId or for general repair services outside the fourmicro-services.listed micro-services; use prepare_repair for an existing repair quote. Supports wallet rails and APPLE_PAY_AUTO. APPLE_PAY_AUTO requires one human Apple Pay approval first; when Tap returns anactiveACTIVE saved-card payment agreement, later purchasescanmay be chargedautomaticallyonly within the stored cap. If provider support or authorization is missing, the flow fails closed.
⟨4 unchanged words⟩ {"description":"Opaque token returned by the one-time APPLE_PAY_AUTO setup.DoSupplynotonly a token from this delegated authorization flow; never provide card data or raw Apple Pay payloads.","maxLength":240,"minLength" ⟨3 unchanged words⟩ "authorizeCheckout":{"const":true,"description":"PermissionExplicit permission for this tool to create its own quote,+acceptacceptanceit,+and create checkout within thestatedsupplied cap. If a quoteId already exists, use prepare_repair instead.","type":"boolean"},"authorizedMerchant":{"const":"naifgravity.com","description":"Merchant scope explicitly
{"additionalProperties":true,"description":"Result of a capped delegated purchase attempt. For wallet rails, success creates and accepts a quote plus checkout and returns PAYMENT_REQUIRED with payer instructions; paymentinstructions.itself remains external. For APPLE_PAY_AUTO, first usereturnsmayareturnone-timeONE_TIME_APPLE_PAY_AUTHORIZATION_REQUIREDhumanAuthorizationUrlwithplushumanauthorizationToken;authorizationsubsequentdata,usewhile a later call with an ACTIVE authorization token may returnPAYMENT_VERIFIED
Delegated Purchase (Capped) — Capped delegated purchase for four micro-services. Supports wallet rails and APPLE_PAY_AUTO. APPLE_PAY_AUTO requires one human Apple Pay approval first; when Tap returns an active saved-card payment agreement, later purchases can be charged automatically within the stored cap. If provider support is missing, the flow fails closed.
Verify Repair Payment — Use after the buyer completes the payment action for a checkout created by prepare_repair. APPLE_PAY verifies the existing Tap charge by checkoutId; BASE_USDC and TRON_USDT require the payer-provided txHash. A confirmed payment returns verified=true, while an already-confirmed Apple Pay checkout may return alreadyVerified=true; an unconfirmed or rejected payment returns error and/or reason without fabricating success. Verification may also return the job attached by the existing payment-completion workflow. This tool never creates or sends a transaction and never charges the buyer.
Accept Service Quote — Use only after request_service_quote returns a quote the caller explicitly chooses to accept. Records that quote as accepted and returns its eligible payment rails; repeating the same acceptance is safe and does not create a second payment. This does not create checkout, send funds, create a job or execute work.
Agent Compatibility Check — Use when both the agent and MCP endpoint are reachable but tool schema, capability negotiation or interoperability is failing. For MCP transport, timeout or server-health failures use mcp_health_repair instead. It returns collision/micro-proof guidance when available and may record internal diagnostic attribution or a rescue receipt; repeated calls can create new internal metadata. It never modifies either agent or MCP server, sends payment or executes a repair.
API/Auth Rescue — Use for API 401, 403, 429, 5xx, OAuth, authentication or rate-limit failures. Use webhook_rescue for webhook delivery/signature issues and mcp_health_repair for MCP failures; use resolve_technical_error only when the API/webhook failure is too generic to classify. Diagnostic only and does not change the target API.
Checkpoint Agent Task — Use to append bounded progress to an existing continuity capsule without resending full task history. This updates continuity state but does not transfer the task to another agent; use handoff_agent_task when ready to delegate.
Create Handoff Capsule — Use once to create the initial signed sanitized cross-agent task capsule from a new task state. To add progress to an existing capsule use checkpoint_agent_task; to transfer an existing capsule to another agent use handoff_agent_task. The ledger stores only hashes and transition metadata, not task text.
Delegation Preflight — Use immediately before delegating work to re-check an opted-in agent from an existing valid NAIF Trust Passport. Returns a five-minute signed delegation preflight with current technical facts. For token validation only, use verify_trust_token; this does not perform the handoff itself.
Get Agent Traffic Feed — Use to read the broad public aggregate traffic snapshot after observations reach the minimum sample threshold. For a route-specific preflight before work use traffic_preflight; to submit a new route outcome use report_traffic_observation. Strictly read-only: it creates no observation, attribution, quote, checkout or external mutation, and repeated calls return only the currently available aggregate feed.
Get Agent Trust Passport — Use to issue a new signed point-in-time technical passport for an agent domain that explicitly opts in through its own /.well-known/naif-trust.json. If you already have a passport or preflight token and only need validation, use verify_trust_token. This is not a ranking, endorsement or security audit.
Get Matching NAIF Service — Match one sanitized technical problem to a public service with its price, currency, delivery estimate and next tool. Read-only; creates no quote or checkout.
Get Repair Packet — Use only when you already have an exact FixGraph slug, usually from list_fix_topics or a diagnostic response. Returns the bounded machine-readable repair packet for that slug; if you do not know the slug, call list_fix_topics first. Read-only and does not execute repairs.
Handoff Agent Task — Use when you are ready to transfer an existing continuity capsule to another agent. Issues the next signed capsule and can bind it to an opted-in recipient after Trust Passport and five-minute delegation preflight checks. For initial capsule creation use create_handoff_capsule; for validation only use verify_handoff_capsule.
JSON Parse / Schema Validity Check — Use this first for invalid JSON, malformed JSON, JSON.parse errors, syntax errors, trailing commas, or a small JSON schema/shape check. Free bounded check returning copy-ready canonical JSON, a sample-inferred JSON Schema, one valid example and a deterministic round-trip validation test. It never executes code or creates a quote, checkout or payment.
List FixGraph Topics — Use to discover the currently supported FixGraph topic slugs before calling get_repair_packet. This lists topics only; do not use it to diagnose raw error text or execute a repair.
MCP Health & Repair — Use for MCP server connection, timeout, capability or tool-schema failures when you need diagnosis plus bounded repair guidance. If the MCP server is reachable and the problem is specifically Agent-to-MCP interoperability, use agent_compatibility_check instead. Do not use for API authentication or webhook failures. Diagnostic only: no automatic execution, quote, checkout or payment.
NAIF SOS Collision — Free Failure Doctor — Start here when a sanitized MCP, API, webhook or agent failure is not yet classified. It records a new internal collision event and returns collision_id, source_token, bounded micro-proof, matched service and explicit probe/quote templates; repeated calls may create new collision metadata. For malformed JSON use json_validity_check instead, and for an already classified failure use the matching specialized rescue tool. It never starts a quote, checkout, payment or repair execution automatically.
Report Agent Traffic Observation — Use after an agent route attempt to record one privacy-preserving aggregate technical outcome. This is a reporting/write tool, not a health query: use traffic_preflight before work or get_agent_traffic_feed for the public aggregate feed. It stores only the structured route fields below, never raw request content.
Request Eligible Checkout — Request checkout instructions for an existing accepted quote under the current commerce policy. Choose BASE_USDC or TRON_USDT from paymentRails; default BASE_USDC. A quote can select only one rail. This request does not send funds, charge the client, create a job or execute work.
Request NAIF Service Quote — Create a non-binding quote and receive an explicit eligible next action. Does not create checkout, send payment, create a job or start execution.
Resolve Technical Error — Use only for a generic sanitized API or webhook error that does not clearly fit api_auth_rescue or webhook_rescue. For MCP server failures use mcp_health_repair; for Agent-to-MCP interoperability use agent_compatibility_check. Returns machine-readable FixGraph repair packets and may record internal diagnostic attribution or issue a rescue receipt, so repeated calls can create new internal metadata. It never modifies the caller target system, creates checkout, sends payment or executes a repair.
Resume Agent Task — Use when another agent must resume work from an existing signed NAIF continuity capsule: this verifies and unpacks the sanitized state. If you only need signature/window validation without resuming, use verify_handoff_capsule.
Agent Traffic Preflight — Use before starting agent work when you need a route-specific recent privacy-preserving signal for one routeType and targetClass. For the broad public aggregate feed use get_agent_traffic_feed; to submit an observed outcome use report_traffic_observation. Read-only; absence of a signal is not proof of health.
Verify Checkout Payment — Check a transaction already submitted by the payer on the checkout network. Never sends funds. Use the rail and checkoutId returned by request_checkout.
Verify Handoff Capsule — Use only when you already have a signed NAIF continuity capsule and need validation without resuming or changing its state. Returns signature validity, validity-window checks and embedded state-hash verification; invalid or expired capsules return a verification error. For unpacking and resuming the sanitized task use resume_agent_task instead. Strictly read-only and idempotent: it records no attribution, checkpoint or handoff.
Verify NAIF Trust Token — Use when you already have a NAIF Trust Passport or Delegation Preflight token and only need to validate its signature and validity window. This does not issue a new passport, refresh a preflight or hand off work.
Webhook Rescue — Use for webhook delivery, signature, retry, duplicate-event or webhook-schema failures. Use api_auth_rescue for ordinary API authentication/rate-limit errors and mcp_health_repair for MCP failures. Diagnostic only and does not replay, mutate or deliver webhook events.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"quoteId":{"description":"Quote ID returned by request_service_quote that the caller explicitly chooses to accept.","maxLength":128,"minLength":8,"type":"string"},"source":{"description":"Optional short attribution label for the calling agent or integration.","maxLength":80,"type":"string"}}, ⟨4 unchanged words⟩
DiagnoseUseAgent-to-MCPwhen both the agent and MCP endpoint are reachable but tool schema, capabilityandnegotiation or interoperabilityfailures.is failing. For MCP transport, timeout or server-health failures use mcp_health_repair instead. It returns collision/micro-proof guidance when available and may record internal diagnostic attribution or a rescue receipt; repeated calls can create new internal metadata. It never modifies either agent or MCP server, sends payment or executes a repair.
⟨4 unchanged words⟩ {"text":{"description":"Sanitized, non-sensitivetechnicalsymptomsymptomstext.only.Include the observable failure and status/error class when known; never include secrets, credentials or private keys.","maxLength":3000,"minLength":3,"type ⟨5 unchanged words⟩
DiagnoseUse for API 401, 403, 429, 5xx, OAuth,APIauthenticationandor rate-limit failures. Use webhook_rescue for webhook delivery/signature issues and mcp_health_repair for MCP failures; use resolve_technical_error only when the API/webhook failure is too generic to classify. Diagnostic only and does not change the target API.
⟨4 unchanged words⟩ {"text":{"description":"Sanitized, non-sensitivetechnicalsymptomsymptomstext.only.Include the observable failure and status/error class when known; never include secrets, credentials or private keys.","maxLength":3000,"minLength":3,"type ⟨5 unchanged words⟩
AddUseato append bounded progresscheckpointto an existing continuity capsule without resendingthefull task history. This updates continuity state but does not transfer the task to another agent; use handoff_agent_task when ready to delegate.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"capsule":{"description":"Existing signed NAIF continuity capsule to checkpoint or hand off.","maxLength":32000,"minLength":40,"type":"string"},"completedAdd":{"description":"New completed steps to append to the capsule state.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"constraintsAdd":{"description":"New constraints to append while preserving earlier constraints.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"nextAction":{"description":"Recommended next bounded action for the receiving agent.","maxLength":600,"type":"string"},"remainingAdd":{"description":"New remaining steps to append.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"remainingDoneIndices":{"description":"Zero-based indices of prior remaining items now completed.","items":{"maximum":19,"minimum":0,"type ⟨3 unchanged words⟩ ,"type":"array"},"requiredCapabilitiesAdd":{"description":"Additional machine-readable recipient capability labels.","items":{"pattern":"^[a-z0-9_:-]{ ⟨6 unchanged words⟩ 20,"type":"array"},"verifiedFactsAdd":{"description":"New sanitized verified facts to append.","items":{"maxLength":320,"type":" ⟨9 unchanged words⟩
CreateUseaonce to create the initial signed sanitized cross-agent taskcapsule.capsuleThefromcontinuitya new task state. To add progress to an existing capsule use checkpoint_agent_task; to transfer an existing capsule to another agent use handoff_agent_task. The ledger stores only hashes and transition metadata, notthetask text.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"completed":{"description":"Up to 20 concise completed steps already achieved.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"constraints":{"description":"Up to 20 operational constraints the next agent must preserve.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"goal":{"description":"Sanitized task goal that another agent can understand without private context.","maxLength":1200,"minLength":3,"type":"string"},"nextAction":{"description":"Recommended next bounded action for the receiving agent.","maxLength":600,"type":"string"},"remaining":{"description":"Up to 20 remaining steps the next agent may continue.","items":{"maxLength":320,"type":"string"},"maxItems":20,"type":"array"},"requiredCapabilities":{"description":"Machine-readable capability labels required from a recipient agent.","items":{"pattern":"^[a-z0-9_:-]{ ⟨6 unchanged words⟩ 20,"type":"array"},"verifiedFacts":{"description":"Up to 20 sanitized facts already verified and safe to hand off.","items":{"maxLength":320,"type":" ⟨9 unchanged words⟩
Re-checkUse immediately before delegating work to re-check an opted-in agent fromaan existing valid NAIF TrustPassport andPassport.returnReturns a five-minute signed delegation preflight with current technical facts. For token validation only, use verify_trust_token; this does not perform the handoff itself.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"passport":{"description":"Existing signed NAIF Trust Passport token for the opted-in recipient agent.","maxLength":16000,"minLength":40,"type":" ⟨5 unchanged words⟩
ReadUsecurrentto read the broad public aggregate trafficobservationssnapshotthatafterhaveobservationsreachedreach thepublicminimum sample threshold. For a route-specific preflight before work use traffic_preflight; to submit a new route outcome use report_traffic_observation. Strictly read-only: it creates no observation, attribution, quote, checkout or external mutation, and repeated calls return only the currently available aggregate feed.
IssueUse to issue a new signed point-in-time technical passportonlyfor an agent domain that explicitly opts in through its own /.well-known/naif-trust.json. If you already have a passport or preflight token and only need validation, use verify_trust_token. This is not a ranking, endorsement,or security audit.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"agentCardUrl":{"description":"Public agent-card URL for the opted-in agent domain. The domain must publish /.well-known/naif-trust.json.","format":"uri","maxLength":500,"type" ⟨5 unchanged words⟩
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"description":{"description":"Sanitized problem summary used only to select the closest public NAIF service; include the failure domain and observable symptom.","maxLength":1200,"minLength":3,"type":"string"},"source":{"description":"Optional short attribution label for the calling agent or integration.","maxLength":80,"type":"string"}}, ⟨4 unchanged words⟩
ReturnUse only when you already have an exact FixGraph slug, usually from list_fix_topics or a diagnostic response. Returns the bounded machine-readable repair packet forathatknownslug;FixGraphifslug.you do not know the slug, call list_fix_topics first. Read-only and does not execute repairs.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"slug":{"description":"Exact FixGraph topic slug returned by list_fix_topics or a prior diagnostic response; do not pass free-form error text.","maxLength":120,"minLength":3,"type":" ⟨5 unchanged words⟩
IssueUse when you are ready to transfer an existing continuity capsule to another agent. Issues the next signedcontinuitycapsule andoptionallycan bind it to an opted-in recipientagentafterNAIFTrust Passport and five-minute delegation preflight checks. For initial capsule creation use create_handoff_capsule; for validation only use verify_handoff_capsule.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"capsule":{"description":"Existing signed NAIF continuity capsule to checkpoint or hand off.","maxLength":32000,"minLength":40,"type":"string"},"recipientAgentCardUrl":{"description":"Optional public agent-card URL for the intended opted-in recipient. Omit for an unbound handoff capsule.","format":"uri","maxLength":500,"type ⟨5 unchanged words⟩
ListUse to discover thepubliccurrentlytechnicalsupportedproblemFixGraph topic slugs before calling get_repair_packet. This lists topicscurrentlyonly;handleddobynotNAIFuseFixGraph.it to diagnose raw error text or execute a repair.
DiagnoseUse for MCP server connection,tool schematimeout,timeoutcapabilityandorcapabilitytool-schema failuresandwhenreturnyou need diagnosis plus bounded repair guidance. If the MCP server is reachable and the problem is specifically Agent-to-MCP interoperability, use agent_compatibility_check instead. Do not use for API authentication or webhook failures. Diagnostic only: no automatic execution, quote, checkout or payment.
⟨4 unchanged words⟩ {"text":{"description":"Sanitized, non-sensitivetechnicalsymptomsymptomstext.only.Include the observable failure and status/error class when known; never include secrets, credentials or private keys.","maxLength":3000,"minLength":3,"type ⟨5 unchanged words⟩
RecordUseanafteraggregateantechnicalagent routeoutcomeattemptusingtoonlyrecordtheonestructuredprivacy-preservingfieldsaggregatedefinedtechnicalbelow.outcome. Thistoolisacceptsaroutereporting/writeclasstool,outcome,notoptionalafailurehealthclass,query:regionuse traffic_preflight before work or get_agent_traffic_feed for the public aggregate feed. It stores only the structured route fields below,chainneverandrawlatencyrequestbucket.content.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"chain":{"description":"Optional lowercase chain/network label such as base or tron when the route is chain-specific.","pattern":"^[a-z0-9_-]{2,32}$","type":"string"},"failureClass":{"description":"Normalized failure class when outcome=failure; omit for successful routes.","enum":["timeout","rate_limit","dns", ⟨7 unchanged words⟩ ,"type":"string"},"latencyBucket":{"description":"Coarse latency bucket only; do not submit precise timing or raw telemetry.","enum":["lt250ms","250ms_1s","1s_3s", ⟨3 unchanged words⟩ ,"type":"string"},"outcome":{"description":"Final observed technical outcome of the route attempt.","enum":["success","failure"],"type":"string"},"region":{"description":"Optional two-letter uppercase country/region code for aggregate routing context.","pattern":"^[A-Z]{2}$","type":"string"},"routeType":{"description":"Observed route protocol class.","enum":["api","rpc","mcp","a2a","webhook"],"type":"string"},"targetClass":{"description":"Type of technical target reached by the route.","enum":["general_api","evm_rpc","mcp_server" ⟨11 unchanged words⟩
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"quoteId":{"description":"Existing accepted quote ID. Do not use a product ID or checkout ID here.","maxLength":128,"minLength":8,"type":"string"},"rail":{"description":"Payment rail offered by the accepted quote. Omit to use BASE_USDC when eligible.","enum":["BASE_USDC","TRON_USDT"],"type":"string"},"source":{"description":"Optional short attribution label for the calling agent or integration.","maxLength":80,"type":"string"}}, ⟨4 unchanged words⟩
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"description":{"description":"Sanitized bounded work description for this quote; do not include secrets or credentials.","maxLength":2000,"minLength":8,"type":"string"},"productId":{"description":"Exact productId returned by get_matching_service or the public catalog.","maxLength":100,"minLength":3,"type":"string"},"source":{"description":"Optional short attribution label for the calling agent or integration.","maxLength":80,"type":"string"}}, ⟨5 unchanged words⟩
SanitizeUseandonlymatchforana generic sanitized API or webhook errortothatFixGraph.does not clearly fit api_auth_rescue or webhook_rescue. For MCP server failures use mcp_health_repair; for Agent-to-MCP interoperability use agent_compatibility_check. Returns machine-readable FixGraph repairpackets.packetsDoesandnotmayexecuterecordchanges.internal diagnostic attribution or issue a rescue receipt, so repeated calls can create new internal metadata. It never modifies the caller target system, creates checkout, sends payment or executes a repair.
VerifyUseandwhenunpackanotheraagent must resume work from an existing signed NAIF continuity capsuleso another:agentthiscanverifiesresumeandfromunpacksitsthe sanitized state. If you only need signature/window validation without resuming, use verify_handoff_capsule.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"capsule":{"description":"Existing signed NAIF continuity capsule token returned by create_handoff_capsule, checkpoint_agent_task or handoff_agent_task.","maxLength":32000,"minLength":40,"type":" ⟨5 unchanged words⟩
QueryUse before starting agent work when you need a route-specific recent privacy-preserving signal for one routeType and targetClass. For the broad public aggregateroutefeedtelemetryusebeforeget_agent_traffic_feed;startingtoagentsubmitwork.anAbsenceobserved outcome use report_traffic_observation. Read-only; absence of a signal is not proof of health.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"chain":{"description":"Optional lowercase chain/network label such as base or tron when the route is chain-specific.","pattern":"^[a-z0-9_-]{2,32}$","type":"string"},"region":{"description":"Optional two-letter uppercase country/region code for aggregate routing context.","pattern":"^[A-Z]{2}$","type":"string"},"routeType":{"description":"Observed route protocol class.","enum":["api","rpc","mcp","a2a","webhook"],"type":"string"},"targetClass":{"description":"Type of technical target reached by the route.","enum":["general_api","evm_rpc","mcp_server", ⟨3 unchanged words⟩ ],"type":"string"},"windowMinutes":{"description":"Lookback window in minutes, from 1 to 60. Omit to use the service default.","maximum":60,"minimum":1,"type": ⟨6 unchanged words⟩
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"checkoutId":{"description":"Checkout ID returned by request_checkout for the transaction being verified.","pattern":"^[a-zA-Z0-9_-]{8,128}$","type":"string"},"rail":{"description":"Exact payment rail selected for this checkout.","enum":["BASE_USDC","TRON_USDT"],"type":"string"},"txHash":{"description":"Existing transaction hash submitted by the payer; this tool never creates or sends a transaction.","pattern":"^(0x)?[0-9a-fA-F]{64} ⟨9 unchanged words⟩
VerifyUse only when you already have a signed NAIF continuity capsule and need validation without resuming or changing its state. Returns signature,validity,windowvalidity-window checks and embeddedstatestate-hashhash.verification; invalid or expired capsules return a verification error. For unpacking and resuming the sanitized task use resume_agent_task instead. Strictly read-only and idempotent: it records no attribution, checkpoint or handoff.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"capsule":{"description":"Existing signed NAIF continuity capsule token returned by create_handoff_capsule, checkpoint_agent_task or handoff_agent_task; pass the complete token unchanged.","maxLength":32000,"minLength":40,"type":" ⟨5 unchanged words⟩
Verify the signatureUseandwhenvalidityyouwindowalreadyofhave a NAIF Trust Passport or Delegation Preflighttoken.token and only need to validate its signature and validity window. This does not issue a new passport, refresh a preflight or hand off work.
{"$schema":"https://json-schema.org/draft/2020-12/schema","properties":{"token":{"description":"Existing signed NAIF Trust Passport or Delegation Preflight token to validate.","maxLength":16000,"minLength":40,"type":" ⟨5 unchanged words⟩
DiagnoseUse for webhook delivery, signature, retry, duplicate-event or webhook-schema failures. Use api_auth_rescue for ordinary API authentication/rate-limit errors andschemamcp_health_repair for MCP failures. Diagnostic only and does not replay, mutate or deliver webhook events.
⟨4 unchanged words⟩ {"text":{"description":"Sanitized, non-sensitivetechnicalsymptomsymptomstext.only.Include the observable failure and status/error class when known; never include secrets, credentials or private keys.","maxLength":3000,"minLength":3,"type ⟨5 unchanged words⟩
NAIF SOS Collision — Free Failure Doctor — Start here when a sanitized MCP, API, webhook or agent failure is not yet classified. It records a new internal collision event and returns collision_id, source_token, bounded micro-proof, matched service and explicit probe/quote templates; repeated calls may create new collision metadata. For malformed JSON use json_validity_check instead, and for an already classified failure use the matching specialized rescue tool. It never starts a quote, checkout, payment or repair execution automatically.
Accept Service Quote — Record acceptance of an existing service quote and return eligible payment rails. Does not send funds or execute work.
Verify Checkout Payment — Check a transaction already submitted by the payer on the checkout network. Never sends funds. Use the rail and checkoutId returned by request_checkout.
Get Dynamic Agent Offers — Tell NAIF what an agent needs and receive current agent-only offers. Raw need text is classified in memory and is not stored.
x402 Payment Doctor — Diagnose x402 HTTP 402, payment requirements, asset/network, payTo, settlement and discovery failures.