Interface: IRun
Defined in: internal/run-store/types.ts:298
The persisted state of one agent run.
Properties
answer?
optionalanswer?:IStoredAnswer
Defined in: internal/run-store/types.ts:351
The checked answer of a run paused for review (D99); present while the pause is a RESULT_REVIEW.
What the settlement had in hand when the run paused, so the decision can finish it as a run that never paused.
Cleared when the run leaves the review. The reviewer's copy is the envelope's result.
chunkIds
chunkIds:
string[]
Defined in: internal/run-store/types.ts:334
The knowledge-base chunks in the request (D36): the citations and the trace annotation of the resumed answer.
createdAt
createdAt:
Date
Defined in: internal/run-store/types.ts:355
When the run was first stored.
id
id:
string
Defined in: internal/run-store/types.ts:300
The run's id (IPausedResponse.runId); kept across the pauses of one run.
messages
messages:
IMessage[]
Defined in: internal/run-store/types.ts:338
LLMClient messages so far, including pending tool calls; their files without bytes, loaded again on resume (D108).
pending?
optionalpending?:TPauseEnvelope[]
Defined in: internal/run-store/types.ts:343
The envelopes of the pause, present while status is PAUSED: one per tool call of the step that waits for a human (D102), or the one RESULT_REVIEW envelope (D99). Never empty while paused.
profileId
profileId:
string
Defined in: internal/run-store/types.ts:308
The profile the run started on; a resume may switch it (D65), the run is saved with the new one.
request
request:
Omit<IModelRequest,"messages"|"outputSchema">
Defined in: internal/run-store/types.ts:315
The request as built for step 1 (system, modelId, caps); messages live separately below. No outputSchema: a zod schema
is code, not data, so resume takes it again from its caller (D36).
skills?
optionalskills?:ISkill[]
Defined in: internal/run-store/types.ts:326
The skills the call brought (IGatewayRunInput.skills, D124), present when it brought any.
A skill is data, so the run keeps it: a resume offers it again, next to the skills of the profile it resumes on.
The profile's skills are not stored: they are read from the profile of the resume, as its tools are (D65). Emptied by
anonymizeByUser.
startTurnCount
startTurnCount:
number
Defined in: internal/run-store/types.ts:332
The thread's turn count before the run's question was stored (D36).
The memory of a resumed run is extracted from the whole run, not from the resumed part.
status
status:
ERunStatus
Defined in: internal/run-store/types.ts:310
Where the run is in its lifecycle (ERunStatus).
steps
steps:
number
Defined in: internal/run-store/types.ts:353
Model calls done so far; the profile's maxStepsPerRun caps it.
templateVersion
templateVersion:
number
Defined in: internal/run-store/types.ts:336
The version of CHAT_MAIN the request was rendered from (D36): the trace annotation of the resumed steps.
tenantId
tenantId:
string
Defined in: internal/run-store/types.ts:302
The owner; every read is filtered by it.
threadId
threadId:
string
Defined in: internal/run-store/types.ts:304
The thread the run answers on; one thread has at most one open run.
toolNames
toolNames:
string[]
Defined in: internal/run-store/types.ts:317
Tool names resolved for this run; re-fetched from ToolRegistry on resume. dyna_aigw_load_skill among them is rebuilt from skills.
updatedAt
updatedAt:
Date
Defined in: internal/run-store/types.ts:357
When the run was last saved or changed status.
userId
userId:
string
Defined in: internal/run-store/types.ts:306
The user who asked; the pseudonym after anonymizeByUser.