Job Records · Reports over the records one agent job leaves behind

Read what a job left behind before your agent makes the next move.

Every job an agent runs leaves a trail: payments in and out, receipts, the rules it ran under, checkouts, payment proofs, the terms it agreed to, the recommendations it gave and the work still open. Job Records runs eight tools over that trail: five checks, two packs and one brief. Your agent sends the records in the call, and each tool returns a signed report with every total, match and flag computed from them.

See the tools Read the contract

Tool access: 0.10 USDC / call
Tools: Eight tools: five checks, two packs and one brief
Connection: /mcp
Result: Signed report

The eight tools.

Each tool reads one kind of job record and answers one question about it. Amounts are added up as exact decimals, so totals match your books to the cent; USD and USDC are read one to one, and every other currency is kept on its own.

`wever_job-cost-brief`

Reads: One job’s money records: payments received for it, payments made during it, compute or model spend and the cost of each failed run, with amount, asset and label, plus the job’s deliverable and how many finished outputs it produced.
The signed report states: Gross in, every cost out itemized with the failed runs on their own line, the net result and margin, and the cost per finished output with the failed-run overhead shown separately. A month of records goes to the expense close pack, and live wallet activity to the wallet statement.
Verdicts: lossbreak_evenprofitable

`wever_job-receipt-pack`

Reads: Every payment in and out and every receipt your agent holds for one job reference, with the counterparty, payment reference and the job each record names.
The signed report states: One pack: each payment matched to the job and to its receipt, totals for money in and out, every record that names a different job, every payment still waiting on a receipt and every receipt that pairs with no payment. A single Wever receipt is verified by receipt verify.
Verdicts: mismatches_foundgaps_foundcomplete

`wever_policy-replay-check`

Reads: A proposed rule change written as structured rules (per-action caps, period caps, allow and block lists and approval thresholds, each optionally limited to some counterparties or services) and a batch of your agent’s past job traces or payment records.
The signed report states: The new rules replayed over the history in time order: per rule, the past actions it would have blocked, the ones it would have sent for approval and the ones that pass unchanged, with counts and every affected record named. One live payment challenge is held against your policy by the payment policy check.
Verdicts: would_blockwould_need_approvalno_change

`wever_duplicate-order-check`

Reads: One checkout that timed out or ended unclear (merchant, item, amount, time and the order reference when known) and the merchant order or receipt records your agent can see, with the time they were exported.
The signed report states: Matching order found, no matching order or indeterminate, with each field compared: order reference, merchant, item, amount, time and status. A matching completed order means a second order would duplicate it; indeterminate names the open point to settle first.
Verdicts: matching_order_foundindeterminateno_matching_order

`wever_payment-recovery-check`

Reads: One settled payment (amount, service or endpoint, transaction reference and time), your agent’s account of the missing result and the service responses it recorded.
The signed report states: Whether the records show the result delivered, not delivered or unresolved, which responses tie to the payment and why, and the next step they support: re-request with the same payment reference, present the payment proof to the seller, or hold the matter as unresolved under the payment already made.
Verdicts: deliverednot_deliveredunresolved

`wever_terms-change-check`

Reads: Two versions of the same terms, policy or agreement: the one in force when your agent bought or signed, and the current one, each with its effective date, and the date of the comparison.
The signed report states: Each changed, added and removed section with a plain label for what it means to the buyer: price or fee, refund or cancellation, liability or dispute, term or renewal, or removed, with a changed figure shown before and after. One document’s evidence is the terms evidence check, and contract redlines are the contract diff check.
Verdicts: buyer_impact_changesother_changes_onlyno_changes

`wever_recommendation-disclosure-check`

Reads: One recommendation your agent is about to give, the commercial terms behind it (an affiliate commission with its rate or amount, a paid placement, a sponsorship, or none) and where it will appear.
The signed report states: Whether the interest is disclosed in the recommendation itself, whether the wording names the right kind of interest and matches the rate or amount, whether it comes before the link or product name, and the exact disclosure line to add, written for that placement.
Verdicts: not_disclosedunderstateddisclosednothing_to_disclose

`wever_job-handoff-pack`

Reads: One half-finished job: its objective, the current draft or work product, completed steps, open questions, files produced, decisions made and what remains.
The signed report states: A pack another agent can resume from: a state summary, the decision log with reasons, the open items in working order, a resume brief and a completeness verdict naming anything the next agent would have to guess. A handover package is reviewed by the handover check, and continuity plans by the continuity check.
Verdicts: not_resumableresumable_with_gapsresumable

How agents and people use it.

Your agent

Keeps the records each job produces: what it was paid, what it spent, the receipts, the checkout attempts, the responses it got back and the state of the work.
Calls the tool that answers its question on https://weverlabs.com/mcp before it acts: before it orders again, pays again, adopts a new rule, gives a recommendation or hands the job to another agent.
Acts on the verdict and keeps the signed report with the job, so the next step starts from a record.

You

See what each job earned once failed runs are counted, with the cost of every finished output.
Try a rule change against last month’s history and see exactly which past actions it would have stopped.
Verify any report against Wever's published trust key whenever you need to show where a number or a verdict came from.

A worked example

What did this job actually make?

The request

{
  "name": "wever_job-cost-brief",
  "arguments": {
    "job_reference": "JOB-42",
    "deliverable": "Twelve vendor research summaries for a procurement team",
    "output_count": 12,
    "output_unit": "summary",
    "asset": "USDC",
    "payments_received": [
      {
        "amount": "450.00",
        "label": "Client payment",
        "reference": "tx-101"
      }
    ],
    "payments_made": [
      {
        "amount": "60.00",
        "label": "Pricing data feed"
      },
      {
        "amount": "25.00",
        "label": "Data for another client",
        "job_reference": "JOB-7"
      }
    ],
    "compute_spend": [
      {
        "amount": "180.40",
        "label": "Model tokens"
      },
      {
        "amount": "34.00",
        "label": "GPU minutes"
      }
    ],
    "failed_runs": [
      {
        "amount": "21.70",
        "label": "Run 3 timed out"
      },
      {
        "amount": "16.50",
        "label": "Run 7 returned an empty file"
      }
    ]
  }
}

What the report returns

Verdict: profitable.

Gross in	450.00 USDC
Payments made	60.00 USDC
Compute and model spend	214.40 USDC
Failed runs	38.20 USDC, 12.22% of spend
Net result	137.40 USDC, 30.53% margin
Cost per summary	26.05 USDC
Failed-run overhead per summary	3.18333333 USDC

Job JOB-42: 450.00 USDC in, 312.60 USDC out (failed runs 38.20 USDC, 12.22% of spend), net 137.40 USDC (30.53% margin); 12 summaries at 26.05 USDC each; 1 record naming another job left out.

The 25.00 USDC payment names job JOB-7, so the brief lists it on its own and keeps it out of this job’s totals. The two failed runs add 3.18 USDC to the cost of every summary; with them counted, each of the twelve summaries cost 26.05 USDC against 37.50 USDC of revenue.

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 Job Records contract lists the schemas, limits, verdicts and signing details. For a month of payment, receipt and refund records, the expense close pack; for live wallet activity, the wallet statement; to verify one receipt, the receipt verify check; for one live payment challenge, a handover package or a continuity plan, the agent preflight checks; and for a single terms document or a contract redline, the DiligenceOps checks. For money records across many jobs, start with Agent Treasury.

Browse all tools