WarrenClient

Defined in: .upstream/src/client/client.ts:70

Public re-exports for the warren-client facade. Internal and external modules can import from here so the file layout under client/ can move without touching call sites.

Constructors

Constructor

new WarrenClient(opts): WarrenClient

Defined in: .upstream/src/client/client.ts:74

Parameters

opts

WarrenClientOptions

Returns

WarrenClient

Properties

config

readonly config: WarrenClientConfig

Defined in: .upstream/src/client/client.ts:71

Methods

cancelPlanRun()

cancelPlanRun(planRunId): Promise<CancelPlanRunResponse>

Defined in: .upstream/src/client/client.ts:345

POST /plan-runs/:id/cancel — flip the plan-run to cancelled and best-effort cancel its in-flight child run. Idempotent: a plan-run already in a terminal state returns { alreadyTerminal: true, cancelledChild: null } without firing a second cancel.

Parameters

planRunId

string

Returns

Promise<CancelPlanRunResponse>

close()

close(): Promise<void>

Defined in: .upstream/src/client/client.ts:394

Returns

Promise<void>

createPlanRun()

createPlanRun(input): Promise<CreatePlanRunResponse>

Defined in: .upstream/src/client/client.ts:310

POST /plan-runs — dispatch a serial plan execution against a seeds plan. Walks the plan’s children one at a time, gating each on the previous PR merging. Re-dispatching the same planId after children land resumes from the next open child (idempotent resume contract).

Parameters

input

CreatePlanRunInput

Returns

Promise<CreatePlanRunResponse>

createProject()

createProject(input): Promise<ProjectRow>

Defined in: .upstream/src/client/client.ts:137

Parameters

input

CreateProjectInput

Returns

Promise<ProjectRow>

createRun()

createRun(input): Promise<SpawnRunResponse>

Defined in: .upstream/src/client/client.ts:186

Parameters

input

CreateRunInput

Returns

Promise<SpawnRunResponse>

dispatch()

dispatch(input): Promise<SpawnRunResponse>

Defined in: .upstream/src/client/client.ts:195

POST /runs with ergonomic warren run field names. Wraps createRun.

Parameters

input

DispatchRunInput

Returns

Promise<SpawnRunResponse>

getAgent()

getAgent(name, query?): Promise<AgentRow>

Defined in: .upstream/src/client/client.ts:174

Parameters

name

string

query?

ListAgentsQuery = {}

Returns

Promise<AgentRow>

getPlanRun()

getPlanRun(planRunId): Promise<PlanRunDetailResponse>

Defined in: .upstream/src/client/client.ts:324

GET /plan-runs/:id — full plan-run status: row + children + fanned-out child RunRow[] (excludes children whose runId is still null) so callers can render a detail view in one round-trip.

Parameters

planRunId

string

Returns

Promise<PlanRunDetailResponse>

getProject()

getProject(projectId): Promise<ProjectRow>

Defined in: .upstream/src/client/client.ts:123

Parameters

projectId

string

Returns

Promise<ProjectRow>

getRun()

getRun(runId): Promise<RunRow>

Defined in: .upstream/src/client/client.ts:231

Parameters

runId

string

Returns

Promise<RunRow>

listAgents()

listAgents(query?): Promise<ListAgentsResponse>

Defined in: .upstream/src/client/client.ts:166

Parameters

query?

ListAgentsQuery = {}

Returns

Promise<ListAgentsResponse>

listPlanRuns()

listPlanRuns(filter?): Promise<ListPlanRunsResponse>

Defined in: .upstream/src/client/client.ts:329

GET /plan-runs[?project=&state=] — list plan-runs, optionally filtered.

Parameters

filter?

ListPlanRunsFilter = {}

Returns

Promise<ListPlanRunsResponse>

listProjects()

listProjects(): Promise<ListProjectsResponse>

Defined in: .upstream/src/client/client.ts:127

Returns

Promise<ListProjectsResponse>

listReadyPlans()

listReadyPlans(projectId, signal?): Promise<ListReadyPlansResponse>

Defined in: .upstream/src/client/client.ts:132

GET /projects/:id/ready-plans — approved, undispatched plans with an open child (warren-7937).

Parameters

projectId

string

signal?

AbortSignal

Returns

Promise<ListReadyPlansResponse>

listRuns()

listRuns(): Promise<ListRunsResponse>

Defined in: .upstream/src/client/client.ts:235

Returns

Promise<ListRunsResponse>

probe()

probe(timeoutMs?): Promise<void>

Defined in: .upstream/src/client/client.ts:88

Hit /healthz with a timeout and convert transport-layer failures into WarrenUnreachableError. /healthz is auth-exempt.

Parameters

timeoutMs?

number = DEFAULT_PROBE_TIMEOUT_MS

Returns

Promise<void>

refreshAgents()

refreshAgents(): Promise<RefreshAgentsResponse>

Defined in: .upstream/src/client/client.ts:182

Returns

Promise<RefreshAgentsResponse>

refreshProject()

refreshProject(projectId, input?): Promise<RefreshProjectResponse>

Defined in: .upstream/src/client/client.ts:145

Parameters

projectId

string

input?

RefreshProjectInput = {}

Returns

Promise<RefreshProjectResponse>

refreshProjectAgents()

refreshProjectAgents(projectId): Promise<RefreshProjectAgentsResult>

Defined in: .upstream/src/client/client.ts:159

Parameters

projectId

string

Returns

Promise<RefreshProjectAgentsResult>

request()

request<T>(path, init?): Promise<T>

Defined in: .upstream/src/client/client.ts:111

Helper to send requests and handle common things like bearer token auth, URL concatenation, and response deserialization.

Type Parameters

T

T

Parameters

path

string

init?

RequestInit = {}

Returns

Promise<T>

steer()

steer(runId, input): Promise<SteerRunResponse>

Defined in: .upstream/src/client/client.ts:220

POST /runs/:id/steer — mid-run steering. Forwards an operator message into the burrow inbox; valid only while the run is non-terminal AND a burrow is attached (else ValidationError). Batch runs get nudges here; src/runs/pause.ts drives the blocking-question pause ↔ resume cycle server-side — no explicit “resume” needed after a steer.

Parameters

runId

string

input

SteerRunInput

Returns

Promise<SteerRunResponse>

streamPlanRunEvents()

streamPlanRunEvents(planRunId, opts?): AsyncGenerator<RunEvent, void, void>

Defined in: .upstream/src/client/client.ts:367

Async iterator over GET /plan-runs/:id/events (NDJSON tail). One yield per child-run event row, parsed as RunEvent. Shares the same wire shape and reader as streamRunEvents; each envelope carries its originating runId discriminator.

follow: true keeps the connection open until the client disconnects or the plan-run reaches a terminal state; without it the server replays the current snapshot then closes.
signal aborts the underlying fetch.

There is no sinceSeq — the endpoint replays the full union snapshot from the top on every (re)open. On reconnect, dedupe client-side by (runId, seq) to drop events already seen.

Parameters

planRunId

string

opts?

StreamPlanRunEventsOptions = {}

Returns

AsyncGenerator<RunEvent, void, void>

streamRunEvents()

streamRunEvents(runId, opts?): AsyncGenerator<RunEvent, void, void>

Defined in: .upstream/src/client/client.ts:274

Async iterator over GET /runs/:id/events (NDJSON tail). One yield per event row, parsed as RunEvent. The wire format is NDJSON, not SSE — each line is a JSON envelope terminated by \n.

follow: true keeps the connection open so new events stream as the run progresses. Without it, the server closes once the current backlog is drained.
sinceSeq replays only events with burrowEventSeq > sinceSeq, matching the server’s ?since= semantics.
signal aborts the underlying fetch so callers can tear the connection down on unmount / cancel.

Malformed lines are dropped silently — the stream is best-effort by design, mirroring the UI consumer in src/ui/src/api/client.ts.

Parameters

runId

string

opts?

StreamRunEventsOptions = {}

Returns

AsyncGenerator<RunEvent, void, void>

waitForPlanRun()

waitForPlanRun(planRunId, opts?): Promise<PlanRunRow>

Defined in: .upstream/src/client/client.ts:383

Poll GET /plan-runs/:id until the plan-run reaches a terminal state (succeeded | failed | cancelled) or the timeout/abort fires. Mirrors waitForRun for serial plan executions.