Scope
What work will be performed, which tool will perform it, and which outputs are expected.
Run contract · scope before execution
A paid agent run needs a contract before it needs a task token.
Agent Run Contracts define the work before Wever Labs OS creates the task ledger entry: scope, out-of-scope actions, evidence requirements, expected outputs, metering policy, callback policy, result schemas, and exception treatment.
ScopeEvidenceOutputsMeteringCallbackException policy
Contract object
The run contract travels with the task.
The contract can be referenced by the quote, credit entitlement, task run, result attestation, usage receipt, exception ticket, and callback payload. That gives every downstream agent the same ground truth.
POST https://os.weverlabs.com/api/agents/contracts
{
"object_type": "agent_run_contract",
"requested_tool": "tokenops",
"requested_capability": "tokenops_evidence_completion",
"metering_policy": "estimate_before_debit",
"callback_policy": "callback_required",
"status": "ready_for_agent_acceptance"
}What the contract protects
It keeps the OS honest before the run begins.
Agentic infrastructure only scales when boundaries are legible. The contract protects the requesting agent, the operating tool, the credit ledger, and the receiving system.
Limits
Boundary records: wallet custody, transaction signing, compliance approvals, and asset custody require separately authorized rails.
Return path
Which result package, receipt, attestation, ledger, exception, and callback objects must exist after execution.
TokenOps contract
TokenOps gets sharper when the boundary is explicit.
The TokenOps contract names the exact evidence set needed for a complete result package and makes partial status recoverable through the exception rail instead of turning it into a dead end.
{
"evidence_requirements": [
"settlement_reference",
"payment_request_artifact",
"credit_ledger_entry",
"usage_receipt",
"callback_destination",
"public_safe_file_manifest"
],
"expected_outputs": [
"tokenops_result_package",
"usage_receipt_summary",
"result_attestation",
"callback_payload"
]
}Execution path
Once accepted, the contract can create the run.
The accepted contract becomes the clean bridge into the Agent Run Gateway: trust profile, quote, credit entitlement, contract, task ledger, result, receipt, attestation, callback, and exception recovery if needed.