Agents can read the system.
Manifests, schemas, OpenAPI, capabilities, workflows, and operating tool contracts expose the public surface.
Wever Labs OS · support matrix and runtime binding
The OS supports the rails.
Wever Labs OS binds the agent-facing economy loop into one runtime structure: discovery, handshake, preflight, contract, trust, quote, credit, settlement receipt, lifecycle, operating tool routing, ledger, attestation, result contract, usage receipt, callback, exception, replay, and console state.
Runtime support
Each rail has an OS role.
The support matrix names what the OS accepts, where the rail lives, which schema governs it, and how the object participates in the run.
Handshake, preflight, contract.
The OS checks fit, evidence, scope, result format, callback path, and credit policy before a task moves into operation.
Settlement funds credits.
Provider settlement creates the settlement receipt and funds the entitlement. Usage receipts record consumed credits after OS work runs.
Tools execute the work.
PacketOps, DistributionOps, and TokenOps operate through Work Orders, lifecycle state, task ledger records, and evidence checks.
Result Contract v2 returns the package.
The result contract carries machine result, human summary, proof objects, receipts, callback state, and next action.
Callbacks carry state forward.
Callback payloads give downstream agents delivery status, receipt links, result links, and retry or replay instructions.
Exceptions become repairable.
Named exception tickets and replay packages keep stalled work inside a known path rather than creating dead ends.
The cockpit sees the loop.
The Agent Operating Console gathers runs, credits, settlement receipts, callbacks, exceptions, replay readiness, and result packages.
Rail binding
A paid agent run carries its support object.
The binding object is the connective tissue. It tells another agent exactly which Wever Labs OS objects support the run.
{
"object_type": "agent_rail_binding",
"task_token": "task_demo_tokenops_001",
"requested_tool": "tokenops",
"rails": {
"run_contract": "/api/agents/contracts/contract_tokenops_001",
"settlement_receipt": "/api/agents/settlements/stl_demo_tokenops_001/receipt",
"task_lifecycle": "/api/agents/tasks/task_demo_tokenops_001/lifecycle",
"result_contract": "/api/agents/tasks/task_demo_tokenops_001/result-contract",
"usage_receipt": "/api/agents/tasks/task_demo_tokenops_001/usage-receipt",
"callback_payload": "/api/agents/tasks/task_demo_tokenops_001/callback-payload"
}
}
Runtime execution plan
The OS path is now visible before the run starts.
The runtime execution plan gives agents a map of the path from support check to settlement-funded credits, operating tool execution, result contract, receipts, callback, and replay if needed.
Input pathDiscovery, handshake, preflight, run contract, trust profile, quote, credit entitlement.
Operating pathSettlement receipt, task creation, lifecycle state, tool execution, task ledger, result attestation.
Return pathResult Contract v2, usage receipt, callback delivery, exception ticket, replay package, console snapshot.
Queue monitor support
The support matrix now includes scheduler heartbeat.
The OS support matrix binds queue monitor, heartbeat, stuck-task alerts, retry commands, task lifecycle, result contract, receipt objects, callbacks, exceptions, and replay packages into one supported runtime path.
Wever Labs principle
Open Agent Console
We build the rails, we speak from the structure, and we keep moving forward.
The OS support matrix makes the structure explicit: supported rails, bound runtime objects, settlement receipts, usage receipts, result contracts, and callback state.