On this page
Manage the lifecycle of autonomous workflow runs: start one from a config file and an intent, list and inspect runs, and abort, pause or resume them.
#orxtra run
Manage autonomous workflow run lifecycle (start, list, show, abort, pause, resume).
#run start
Start a new autonomous workflow run from a TOML configuration file and a natural-language intent. Opens a pool against the trace database, verifies the schema, dispatches start_run under an operator identity, and prints the new run's UUID on stdout so later run, trace and inbox commands can address it.
Effect: mutating
#Flags
| Name | Short | Type | Default | Env | Description |
|---|---|---|---|---|---|
--config | str | Filesystem path to the run configuration TOML file. | |||
--intent | str | Natural-language description of what the run should accomplish. |
#run list
List every run recorded in the trace database, newest first, with the identifying and status fields the storage layer keeps for each one. Output honours the global --format flag, so the same listing renders as a human-readable table or as JSON for a script or agent to consume directly.
Effect: read_only
#run show
Display the full status report the trace store holds for one run: its current state, the intent it was started with, and the accounting the storage layer keeps alongside it. Exits non-zero with a clear message when no run carries the given identifier. Honours the global --format flag for table or JSON output.
Effect: read_only
#Arguments
| Name | Required | Description |
|---|---|---|
run_id | yes | Unique identifier of the run to display (UUID format). |
#run abort
Send an abort signal to a currently executing workflow run, telling the scheduler to stop dispatching further tasks for it. The signal travels through the trace store's run-control channel rather than through dispatch, so a scheduler running in another process picks it up. Confirms on stdout unless --quiet.
Effect: mutating
#Arguments
| Name | Required | Description |
|---|---|---|
run_id | yes | Unique identifier of the run to abort (UUID format). |
#run pause
Suspend a running workflow, telling the scheduler to stop starting new tasks for it while leaving the run and its task tree intact so it can be resumed later. The signal travels through the trace store's run-control channel, so a scheduler in another process observes it. Confirms on stdout unless --quiet.
Effect: mutating
#Arguments
| Name | Required | Description |
|---|---|---|
run_id | yes | Unique identifier of the run to pause (UUID format). |
#run resume
Restart task execution for a workflow that was previously paused, telling the scheduler to begin dispatching its ready tasks again from the state the pause left behind. The signal travels through the trace store's run-control channel, so a scheduler in another process observes it. Confirms on stdout unless --quiet.
Effect: mutating
#Arguments
| Name | Required | Description |
|---|---|---|
run_id | yes | Unique identifier of the run to resume (UUID format). |