Stats
Actions
Tags
From aux4-agent-skills
Run and manage background shell commands — start, monitor, tail output, and kill jobs using aux4 jobs.
How this skill is triggered — by the user, by Claude, or both
Slash command
/aux4-agent-skills:jobsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Manage background shell commands: $ARGUMENTS
Manage background shell commands: $ARGUMENTS
| Command | Description |
|---|---|
aux4 jobs run "<command>" | Run a command in the background, returns job JSON |
aux4 jobs list | List all jobs |
aux4 jobs list --state RUNNING | List jobs filtered by state |
aux4 jobs status "<id>" | Show job status, exit code, duration |
aux4 jobs output "<id>" | Print captured stdout |
aux4 jobs output "<id>" --stream stderr | Print captured stderr |
aux4 jobs kill "<id>" | Kill a running job |
aux4 jobs killall | Kill all running jobs |
aux4 jobs on "<id>" --success "<cmd>" --failure "<cmd>" --complete "<cmd>" | Register callbacks |
| State | Meaning |
|---|---|
RUNNING | Job is currently executing |
COMPLETED | Job finished with exit code 0 |
FAILED | Job finished with non-zero exit code |
KILLED | Job was manually killed |
When callbacks fire, these variables are available:
AUX4_JOB_ID — Job IDAUX4_JOB_STATE — Final stateAUX4_JOB_EXIT_CODE — Exit codeAUX4_JOB_COMMAND — Original commandAUX4_JOB_DIR — Working directoryUse aux4 jobs run proactively for:
aux4 jobs run "npm run build"aux4 jobs run "kubectl apply -f deploy.yaml"aux4 jobs run "python process.py --input data.csv"aux4 jobs run "aux4 test run"aux4 jobs run "npm run build"
# ... do other work ...
aux4 jobs status "<id>"
aux4 jobs output "<id>"
aux4 jobs run "deploy.sh" --onSuccess "echo deployed" --onFailure "echo FAILED"
aux4 jobs run "terraform apply -auto-approve"
aux4 poll --command "aux4 jobs status <id> | grep state" --expectation "*COMPLETED*" --interval 10 --maxWait 300
npx claudepluginhub aux4/agent-skills --plugin aux4-agent-skillsCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.