Agent Mesh — workers, grains and sagas

Generated from the architecture model

This page is rendered from model/**/*.yaml by tools/docs_model during mkdocs build. Edit the YAML, not this page; hand-written design lives on the linked service, platform and plane pages.

Workers

Worker	Service	Trigger	Output events	Retry	Idempotency key	Concurrency
LeaseHeartbeat	AgentRuntimeService	timer every 15s	AttemptPhaseChanged	fixed ×3	leaseId + renewalSequence	1 by AttemptId
OrphanedAttemptReconciler	AgentRuntimeService	schedule */2 * * * *	AttemptOutcomeReported	exponential ×5	attemptId + outcomeHash	2 by TenantId
CheckpointCompactor	AgentRuntimeService	schedule 0 3 * * *	—	fixed ×3	checkpointId	1 by TenantId
ReleaseCacheWarmer	AgentRuntimeService	event AgentRegistryService.AssetReleasePublished	ReleaseLoaded, ReleaseLoadRejected	exponential ×5	releaseId + adapterVersion	2 by TenantId
TimerScheduler	WorkflowEngineService	reminder per timer dueAt	TimerFired, StepReady	exponential ×5	timerId	1 by RunId
StuckRunReconciler	WorkflowEngineService	schedule */5 * * * *	StepReady, StepRunStateChanged	fixed ×3	stepRunId + lastOutcomeSequence	2 by TenantId
DefinitionCacheWarmer	WorkflowEngineService	event AgentRegistryService.AssetReleasePublished	WorkflowDefinitionLoaded	exponential ×5	releaseId	2 by TenantId

• LeaseHeartbeat (AgentRuntimeService): Renews the lease of every running attempt before expiry through Run Control; a failed renewal pauses the attempt at its next checkpoint.
• OrphanedAttemptReconciler (AgentRuntimeService): Finds attempts whose grain is inactive and whose lease is gone; settles them as Uncertain when a call was in flight, otherwise Cancelled, and reports.
• CheckpointCompactor (AgentRuntimeService): Deletes superseded adapter checkpoints of terminal attempts after the retention window, keeping the last one and every checkpoint referenced by evidence.
• ReleaseCacheWarmer (AgentRuntimeService): Verifies newly published releases against this worker's adapter before the first dispatch.
• TimerScheduler (WorkflowEngineService): Registers an Orleans reminder for every timer entry and applies the fired timer to the run.
• StuckRunReconciler (WorkflowEngineService): Finds workflow runs whose active step runs have no corresponding open attempt or approval in Run Control and re-emits StepReady or fails the step per policy.
• DefinitionCacheWarmer (WorkflowEngineService): Validates and caches newly published workflow definitions.

Grains (actors)

Grain	Service	Key	Reminders / timers	Reentrant	Summary
AttemptGrain	AgentRuntimeService	guid = attemptId	call-timeout, attempt-budget	no	Serialises the loop of one attempt; holds the adapter session, writes checkpoints, enforces the fencing token and the limits, and reports the outcome once.
WorkflowRunGrain	WorkflowEngineService	guid = runId	timer, stuck-detection	no	Serialises Advance for one run, holds the cursor and active step runs, registers timer reminders and commits step-run rows with the outbox in one transaction.

Sagas

None modelled

No sagas are modelled for this item yet.