Runs and invocations

Inspect runs, artifacts, Function invocations, and App invocations.

Executions return a run_id immediately. Use the runs and invocation clients to load the audit trail after the execution finishes.

Runs

runs.ts

const run = await harbor.runtime.execute({
  code: 'return { ok: true }',
})

const detail = await harbor.runs.get({ run_id: run.run_id })
const graph = await harbor.runs.graph({ run_id: run.run_id })
const artifacts = await harbor.runs.listArtifacts({ run_id: run.run_id })
const events = await harbor.runs.events({ run_id: run.run_id })

Run records are workspace-scoped. They are the bridge between an SDK call and the dashboard trace.

Function invocations

Functions are the product name for reusable Orbit Jobs. The SDK keeps the code name visible in the resource group:

function-invocations.ts

const functions = await harbor.jobs.list({ limit: 25 })

const invocation = await harbor.jobs.run({
  job_id: 'job_...',
  input: { dryRun: true },
})

const invocations = await harbor.jobs.invocations.list({
  job_id: 'job_...',
  limit: 25,
})

const invocationDetail = await harbor.jobs.invocations.get({
  invocation_id: 'invocation_...',
})

App invocations

Orbit Apps also expose invocation records:

app-invocations.ts

const apps = await harbor.apps.list({ limit: 25 })

const activity = await harbor.apps.activity.list({
  app_id: 'app_...',
  limit: 25,
})

const invocations = await harbor.apps.invocations.list({
  app_id: 'app_...',
  limit: 25,
})

Use run and invocation records for support, audit, replay, and UI links. Do not depend on terminal output or local logs as the only execution proof.

Runs and invocations

Runs

Function invocations

App invocations

On this page