Agent Run
Custom object AgentRun__c. Tracks the durable lifecycle and aggregate state of one agent execution.
| Field | API name | Type | Description |
|---|---|---|---|
| Agent Definition | AgentDefinition__c | Lookup | System-maintained Agent Definition for this durable agent run. |
| Agent Execution Identity | ExecutionIdentity__c | Lookup | The Agent Execution Identity frozen at run start; blank when the run acts as the user who started it. |
| Anchor Record ID | AnchorRecordId__c | Text | Optional Salesforce record ID that identifies the primary durable business context for this run. |
| Attended By | AttendedBy__c | Lookup | The person a Delegated Principal run works for, captured when that person started it in their own request (a Lightning page, screen flow, or API call). Record data the run’s tools and context return, and records its standard tools change, are limited to what this person can read and edit. Blank for scheduled, triggered, event, and background starts, and for Initiating User runs, which already run as that person. |
| Cancellation Reason | CancellationReason__c | LongTextArea | Bounded reason supplied by the authorized caller requesting cancellation. |
| Cancellation Requested At | CancellationRequestedAt__c | DateTime | Time an authorized caller requested cancellation while a command was already leased. |
| Command Sequence | CommandSequence__c | Number | Monotonic per-run command counter that identifies each durable operation independently of run revision. |
| Completed At | CompletedAt__c | DateTime | System-maintained Completed At for this durable agent run. |
| Configuration Hash | ConfigurationHash__c | Text | SHA-256 hash of the executable agent configuration (definition, active tools, and context providers) that the run last used. A different hash on a later call records a Configuration Changed event. |
| Continuation Mode | ContinuationMode__c | Picklist | System-maintained Continuation Mode for this durable agent run. Values: Model, Conversation, Reinvoke Tool. |
| Current Command | CurrentCommand__c | Lookup | System-maintained Current Command for this durable agent run. |
| Delete After | DeleteAfter__c | DateTime | System-maintained earliest time at which this payload-purged terminal run and its evidence may be permanently deleted. |
| Error Code | ErrorCode__c | Text | System-maintained Error Code for this durable agent run. |
| Error Message | ErrorMessage__c | LongTextArea | System-maintained Error Message for this durable agent run. |
| Execution Context Status | ExecutionContextStatus__c | Picklist | System-maintained Execution Context Status for this durable agent run. Values: Pending, Established. |
| Expected Principal User | ExpectedPrincipalUser__c | Lookup | Allowlisted Delegated Principal user snapshotted at run start. Loopback dispatch may resolve only this user. |
| External Correlation Key | ExternalCorrelationKey__c | Text | Caller-supplied business key used to correlate and idempotently start a run within one agent definition. |
| Initiating User | InitiatingUser__c | Lookup | System-maintained Initiating User for this durable agent run. |
| Input | Input__c | LongTextArea | System-maintained Input for this durable agent run. |
| Input Hash | InputHash__c | Text | System-maintained Input Hash for this durable agent run. |
| Input Token Count | InputTokenCount__c | Number | System-maintained Input Token Count for this durable agent run. |
| Invocation Entry Point | InvocationEntryPoint__c | Picklist | System-maintained Invocation Entry Point for this durable agent run. Values: Apex, Flow, REST, Platform Event, Chat. |
| Last Activity At | LastActivityAt__c | DateTime | System-maintained Last Activity At for this durable agent run. |
| Last Event Sequence | LastEventSequence__c | Number | Monotonic sequence allocated while the run row is locked for the append-only event ledger. |
| Last Lifecycle Publish Failure At | LastLifecyclePublishFailureAt__c | DateTime | Most recent time Core observed a lifecycle platform-event publish failure for this run. |
| Lifecycle Publish Failure Count | LifecyclePublishFailureCount__c | Number | Payload-free count of best-effort lifecycle platform-event publish failures for this run. |
| Loopback Named Credential | LoopbackNamedCredential__c | Text | System-maintained Loopback Named Credential for this durable agent run. |
| Max Retries | MaxRetries__c | Number | System-maintained retry budget captured when this durable run started. |
| Model Call Count | ModelCallCount__c | Number | System-maintained Model Call Count for this durable agent run. |
| Model Calls Before Reply | ReplyModelCallBase__c | Number | Model Call Count when the person’s latest reply arrived. A conversational run’s model-call budget counts only the calls after it. |
| Model Duration Milliseconds | ModelDurationMs__c | Number | System-maintained Model Duration Ms for this durable agent run. |
| Output | Output__c | LongTextArea | System-maintained Output for this durable agent run. |
| Output Hash | OutputHash__c | Text | System-maintained Output Hash for this durable agent run. |
| Output Token Count | OutputTokenCount__c | Number | System-maintained Output Token Count for this durable agent run. |
| Parent Run | ParentRun__c | Lookup | Optional lineage link to another Pluto Core run. Parent and child do not share budget, transcript, identity, or lifecycle. |
| Payload Purged At | PayloadPurgedAt__c | DateTime | System-maintained Payload Purged At for this durable agent run. |
| Payload Retention Days | PayloadRetentionDays__c | Number | System-maintained Payload Retention Days for this durable agent run. |
| Payload Retention Mode | PayloadRetentionMode__c | Picklist | System-maintained Payload Retention Mode for this durable agent run. Values: Redacted, Full, Hash Only, None. |
| Purge After | PurgeAfter__c | DateTime | Queryable deadline after which terminal run payloads are eligible for purge. |
| Record Retention Days | RecordRetentionDays__c | Number | System-maintained evidence-record retention window captured when this durable run started. |
| Recovery Check At | RecoveryCheckAt__c | DateTime | When scheduled recovery may next examine this run. Core sets it after a recovery step fails for this run so one failing run cannot stall recovery of others. |
| Resolved Principal User | ResolvedPrincipalUser__c | Lookup | System-maintained Resolved Principal User for this durable agent run. |
| Resume JSON Schema | ResumeJsonSchema__c | LongTextArea | System-maintained Resume Json Schema for this durable agent run. |
| Retained Payload Characters | RetainedPayloadCharacters__c | Number | System-maintained Retained Payload Characters for this durable agent run. |
| Retry Count | RetryCount__c | Number | System-maintained Retry Count for this durable agent run. |
| Revision | Revision__c | Number | System-maintained Revision for this durable agent run. |
| Run Duration Milliseconds | RunDurationMs__c | Number | Elapsed wall-clock milliseconds from run start to terminal completion; calculated only for completed runs. |
| Sensitive JSON Keys | SensitiveJsonKeys__c | LongTextArea | System-maintained Sensitive Json Keys for this durable agent run. |
| Start Deduplication Key | StartDeduplicationKey__c | Text | System-generated SHA-256 key that uniquely scopes an external correlation key to one agent definition. |
| Start Request Hash | StartRequestHash__c | Text | SHA-256 fingerprint of the immutable values supplied by the original correlated start request. |
| Started At | StartedAt__c | DateTime | System-maintained Started At for this durable agent run. |
| Status | Status__c | Picklist | System-maintained Status for this durable agent run. Values: Running, Waiting, Paused, Succeeded, Failed, Cancelled. |
| Tokens Before Reply | ReplyTokenBase__c | Number | Total Token Count when the person’s latest reply arrived. A conversational run’s token budget counts only the tokens after it. |
| Tool Call Count | ToolCallCount__c | Number | System-maintained Tool Call Count for this durable agent run. |
| Tool Calls Before Reply | ReplyToolCallBase__c | Number | Tool Call Count when the person’s latest reply arrived. A conversational run’s tool-call budget counts only the calls after it. |
| Tool Duration Milliseconds | ToolDurationMs__c | Number | System-maintained Tool Duration Ms for this durable agent run. |
| Total Token Count | TotalTokenCount__c | Number | System-maintained Total Token Count for this durable agent run. |
| Unmetered Model Attempts | UnmeteredModelAttempts__c | Number | Model attempts that ended without reported token usage, such as a timeout. Core estimates their input tokens from the request size and counts them toward Max Total Tokens. |
| Wait Expires At | WaitExpiresAt__c | DateTime | System-maintained Wait Expires At for this durable agent run. |
| Waiting Command | WaitingCommand__c | Lookup | System-maintained Waiting Command for this durable agent run. |
| Waiting Reason | WaitingReason__c | Text | System-maintained Waiting Reason for this durable agent run. |
Pluto is licensed under the GNU Affero General Public License, version 3 only.