Skip to main content
Every time Pylon runs an AI agent it creates a job — a record with a unique ID, status, and timestamps. These commands let you inspect what happened, debug failures, and manage running containers.

​
pylon jobs

Lists recent jobs across all pipelines (or filtered to one).
Example output:
Job statuses:

​
pylon logs

Streams Docker logs for a job’s container.
​
job-id
string
required
The job ID shown in pylon jobs output. You can use the short 8-character prefix.
If the container no longer exists, Pylon prints the workspace directory path where any saved output may be found.

​
pylon attach

Opens an interactive shell inside a running agent container.
This runs docker exec -it <container> /bin/sh. Useful for inspecting the workspace, environment variables, or agent state mid-run.
The container must still be running. Use pylon jobs to check the status first.

​
pylon kill

Forcibly stops a running agent container.
Equivalent to docker kill <container>. Use this when an agent is stuck or has exceeded its expected runtime.

​
pylon retry

Re-triggers a failed job.
For webhook-triggered pipelines, retry is available by re-sending the original webhook or using pylon test. For cron-triggered pipelines, wait for the next scheduled run.

​
Job storage

Each pipeline stores its job history in a SQLite database at:
Jobs are kept indefinitely. To clear history, use pylon destroy <name> and recreate the pipeline.