Agent Reliability · Checks on whether a service’s reliability records agree with each other and with what it advertises

Every reliability record, read against the records behind it.

Reliability records hold up when the published uptime figure is the one the downtime log supports; the error budget follows from the SLO it cites; the failover arrangement delivers the standby, recovery objectives and triggers the terms claim; the dependency disclosure names the providers, dependent functions and fallbacks the listing states; the degraded behavior on record is the behavior advertised; the incident summary buyers see carries every incident in the history with its duration and cause; maintenance follows the notice, length and frequency the policy sets; and the periodic report’s headlines are the figures its records hold. Agent Reliability reads the records an agent or its principal supplies, checks each one for consistency against the document behind it, and returns a signed verdict with the evidence read and what would change it.

Tool access
0.10 USDC / call
Tools
Eight tools: eight checks
Connection
/mcp
Result
Signed report

The eight tools.

Each tool reads one pair of reliability records for consistency between documents and decides from the material in the call: the uptime report with its downtime log, the error budget record with its SLO, the failover arrangement with the continuity claims, the dependency disclosure with the listing, the degradation record with the advertised behavior, the incident history with the buyer disclosure, the maintenance record with the policy, and the reliability report with its records go in, and a verdict comes out. The report names the figure, incident, provider, mode, window or date behind every finding, the fix for each, and what would change the verdict. Each check compares the records as written; live measurement stays with the endpoint probes, service credits with the credit check, and the judgment of how reliable a service is with the people relying on it.

wever_rely-uptime-check

Reads
A published uptime report with its period, availability, downtime minutes and incident count, beside the incident and downtime log for the same period with each entry’s start, end or minutes.
The signed report states
An availability above what the log supports or not following from the report’s own downtime, fewer downtime minutes than the log records and fewer incidents than it holds, then a lower availability, more downtime or incidents, another period or service and entries outside the period, with availability recomputed from the log.
Price
0.10 USDC per call
Verdicts
rely_uptime_misreportedrely_uptime_partialrely_uptime_unverifiedrely_uptime_faithful

wever_rely-error-budget-check

Reads
An error budget record with the SLO it cites, its target and window, and the budget, minutes consumed, minutes remaining and burn it states, beside the SLO with its target and window.
The signed report states
A target or window unlike the SLO, a budget above what the SLO allows, more budget remaining and a lower burn than the arithmetic gives, then a lower budget, less remaining, a higher burn, another cited SLO and another service, with the budget recomputed from the SLO.
Price
0.10 USDC per call
Verdicts
rely_error-budget_divergentrely_error-budget_partialrely_error-budget_unverifiedrely_error-budget_consistent

wever_rely-failover-check

Reads
A failover arrangement record with its standby, recovery time and recovery point objectives, failover conditions and activation, beside the continuity claims the terms or listing make.
The signed report states
A standby weaker than claimed, a longer recovery time or recovery point, a claimed failover condition the arrangement does not hold and a manual failover where the claim is automatic, then a stronger standby, shorter objectives, unclaimed conditions and unclaimed automatic failover.
Price
0.10 USDC per call
Verdicts
rely_failover_divergentrely_failover_partialrely_failover_unverifiedrely_failover_consistent

wever_rely-dependency-check

Reads
A dependency disclosure with each upstream provider, the functions that depend on it and its fallback, beside the providers, functions and fallbacks the listing or terms state.
The signed report states
A provider, dependent function or fallback the listing states that the disclosure leaves out or contradicts, then a disclosed provider, function or fallback the listing leaves out and another service, provider by provider.
Price
0.10 USDC per call
Verdicts
rely_dependency_discrepantrely_dependency_partialrely_dependency_unverifiedrely_dependency_consistent

wever_rely-degradation-check

Reads
A degradation behavior record with its degraded modes, stale-data limit, write queueing and queue hold, and the functions that stop entirely, beside the behavior the service advertises for partial failure.
The signed report states
An advertised mode the record lacks, data stale longer than advertised, writes refused or held for less time than advertised and a function stopped that is not advertised to stop, then unadvertised modes or queueing, a shorter stale limit, a longer hold and an advertised stop the record keeps running.
Price
0.10 USDC per call
Verdicts
rely_degradation_divergentrely_degradation_partialrely_degradation_unverifiedrely_degradation_consistent

wever_rely-incident-history-check

Reads
An incident history for a period with each incident’s title, duration and cause, beside the summary disclosed to buyers with its incidents, count and total minutes.
The signed report states
Every history incident missing from the disclosure by name, a shorter duration, another cause and a count or total below the history, then a disclosed incident not in the history, a longer duration, a withheld cause, another title and a higher count or total.
Price
0.10 USDC per call
Verdicts
rely_incident-history_misreportedrely_incident-history_partialrely_incident-history_unverifiedrely_incident-history_faithful

wever_rely-maintenance-check

Reads
A maintenance record for a period with each window’s announcement, start, end or minutes and how maintenance time was treated in the uptime figures, beside the maintenance policy’s notice, longest window, monthly limit and treatment.
The signed report states
A window announced with less notice than the policy requires, a window longer than it allows, a month with more windows than it allows and maintenance excluded from uptime where the policy counts it, then maintenance counted where the policy excludes it and windows outside the period.
Price
0.10 USDC per call
Verdicts
rely_maintenance_divergentrely_maintenance_partialrely_maintenance_unverifiedrely_maintenance_consistent

wever_rely-report-check

Reads
A periodic reliability report written for buyers with its headline availability, incident count and maintenance total, beside the uptime report, incident history and maintenance record it summarizes.
The signed report states
A headline availability above the uptime report, fewer incidents than the history holds and less maintenance than the maintenance record holds, then a lower availability, more incidents or maintenance than the records hold and a record for another period.
Price
0.10 USDC per call
Verdicts
rely_report_misreportedrely_report_partialrely_report_unverifiedrely_report_faithful

Start with the two free tools.

On https://weverlabs.com/mcp-free, anonymous and with no wallet, two Agent Reliability tools map the family and show the uptime check at work on synthetic data. Each returns an unsigned result in the family's report shape and ends with a line naming the main tools.

wever_rely-snapshot

Reads
Nothing, or the situation in hand, such as publishing uptime, planning failover or writing a reliability report, and optionally the names of the documents in hand.
Returns
The eight reliability record checks with what each reads, its four verdicts and when each fits, with the checks that fit the situation listed first.
Access
Free on /mcp-free

wever_rely-uptime-sample

Reads
Nothing, or one synthetic service: a search API, a file storage service or a messaging service.
Returns
A synthetic uptime report faithful to its downtime log beside a misreported one, both read by the uptime check’s own reading, with the availability, the downtime, the incidents and the period annotated with why they pass or fail.
Access
Free on /mcp-free

How agents and people use it.

Your agents

  • Call wever_rely-snapshot with the situation in hand to see which checks fit first, before reading any record.
  • Call the check that answers the question on https://weverlabs.com/mcp: when an uptime figure is published, when an error budget is reported, when continuity is claimed, when dependencies are disclosed, when degraded behavior is advertised, when incidents are summarized, when maintenance is scheduled and when a reliability report goes to buyers.
  • Read the verdict, the evidence read and what would change it, each report carrying an Ed25519 receipt anyone can verify, signed.
  • Act on the verdict and its findings: ask for a corrected report, record, disclosure or summary, the records behind it, or carry on as checked.

You

  • See each uptime figure, error budget and failover claim read against the log, SLO or arrangement behind it, with the figure on each side.
  • Read every dependency disclosure, degradation record, incident summary, maintenance record and reliability report against the listing, advertised behavior, history, policy or records behind it.
  • Verify any check against Wever's published trust key whenever you need to show where a verdict came from.

A worked example

One uptime report, read against the downtime log behind it.

The request

{
  "name": "wever_rely-uptime-check",
  "arguments": {
    "report": {
      "service": "Harbor Search API",
      "period_from": "2026-10-01",
      "period_to": "2026-10-31",
      "availability_percent": 99.95,
      "downtime_minutes": 22,
      "incident_count": 3
    },
    "log": {
      "service": "Harbor Search API",
      "period_from": "2026-10-01",
      "period_to": "2026-10-31",
      "entries": [
        {
          "id": "SR-1",
          "started_at": "2026-10-03T02:00:00Z",
          "ended_at": "2026-10-03T02:25:00Z"
        },
        {
          "id": "SR-2",
          "started_at": "2026-10-17T14:10:00Z",
          "minutes": 12
        },
        {
          "id": "SR-3",
          "started_at": "2026-10-28T08:00:00Z",
          "ended_at": "2026-10-28T08:08:00Z"
        }
      ]
    }
  }
}

What the report returns

Verdict: rely_uptime_misreported.

Finding, availability overstatedThe report publishes 99.95% availability where the log’s 45 minutes of downtime over 2026-10-01 to 2026-10-31 leaves 99.899%.
Finding, downtime understatedThe report states 22 minutes of downtime where the log records 45 minutes.
Fix, availability overstatedPublish the availability the log supports for the period, or record the downtime the log leaves out.
Fix, downtime understatedState the downtime minutes the log records for the period.

The uptime report misreports the log: availability overstated and downtime understated.

The log records 45 minutes of downtime across three incidents in October, which leaves 99.899% availability over the month’s 44,640 minutes; the report publishes 99.95% and 22 minutes. The incident count, the period and the service match the log, and the report’s own 22 minutes would give 99.95%, so the report is consistent with itself and not with its log. The report states that publishing the figure the log supports and its downtime moves the verdict to rely_uptime_faithful.

Call it from your agent.

Connect to https://weverlabs.com/mcp and call any of the eight tools. Each call costs 0.10 USDC on Base. Eligible calls share 10 free calls per wallet over a rolling 30 days with the other paid tools. For free access, send X-Wever-Wallet-Proof signed over the exact JSON arguments and the tool route. The payment challenge supplies the x402 terms for a paid call.

Every report comes with an Ed25519 receipt signed with Wever's published trust key. The Agent Reliability contract lists the schemas, limits, verdicts and signing details. To probe one live MCP or x402 endpoint, Endpoint reliability; to restate one SLA, status page or incident on its own, Diligence checks; for your own incident plan, recovery arrangements and an operator’s continuity, Preflight; for retry, rollback and failure-notice designs, Agent Recovery; for an agent’s own traces, logs and summaries, Agent Observability; and for SLA terms and deadlines kept on record, Agent SLA Rail.

Browse all tools