sem-ai-cli

name: sem-ai-cli description: Reference Semaphore's sem-ai CLI command set, command families, output formats, safety defaults, and legacy sem replacement patterns.

sem-ai CLI

Use this skill when you need the exact sem-ai command surface before touching Semaphore state. It is a reference skill, not the incident workflow itself. For active CI triage, pair it with semaphore-debugging.

Core Model

sem-ai is the agent-oriented Semaphore CLI. It uses structured output by default, supports JSON/table/YAML formatting, and shares the same ~/.sem.yaml context file as the legacy sem CLI.

Setup

sem-ai connect simplepractice.semaphoreci.com "$SEMAPHORE_API_TOKEN"
sem-ai context show
sem-ai version --check

Use sem-ai discover when you need a machine-readable command map.

Command Families

# Context and auth
sem-ai connect <host> <token>
sem-ai context list
sem-ai context show
sem-ai context switch <name>

# Project and CI state
sem-ai project list
sem-ai status --project <project> --branch <branch>
sem-ai health --project <project>
sem-ai workflow list --project <project>
sem-ai workflow show <workflow-id>
sem-ai pipeline list --project <project>
sem-ai pipeline show <pipeline-id>
sem-ai job list
sem-ai job show <job-id>

# Logs and failure analysis
sem-ai diagnose <workflow-id>
sem-ai troubleshoot workflow <workflow-id>
sem-ai job log <job-id>
sem-ai test summary --pipeline <pipeline-id>
sem-ai test report --pipeline <pipeline-id>
sem-ai blast-radius <pipeline-id>
sem-ai critical-path <pipeline-id>
sem-ai pipeline topology <pipeline-id>

# Recovery and mutation
sem-ai job stop <job-id>
sem-ai pipeline stop <pipeline-id>
sem-ai pipeline rebuild <pipeline-id>
sem-ai workflow rerun <workflow-id>
sem-ai rerun-failed <pipeline-id>
sem-ai watch <workflow-id>

# Deploys and operations
sem-ai pipeline promote <pipeline-id> --target <promotion-name> --confirm
sem-ai promote-and-wait <pipeline-id> --target <promotion-name> --confirm
sem-ai deploy targets --project <project>
sem-ai secret list
sem-ai notification list
sem-ai task list --project <project>
sem-ai artifact list --scope jobs --id <job-id>
sem-ai artifact get --scope jobs --id <job-id> --path <path>
sem-ai testbox warmup --project <project>
sem-ai testbox run --id <testbox-id> "<command>"
sem-ai testbox ssh --id <testbox-id>
sem-ai testbox stop --id <testbox-id>
sem-ai yaml validate --file .semaphore/semaphore.yml

Output

sem-ai pipeline show <pipeline-id> --format json
sem-ai pipeline show <pipeline-id> --format table
sem-ai pipeline show <pipeline-id> --format yaml
sem-ai --verbose pipeline show <pipeline-id>

Default to JSON for automation. Use table output only when a human needs to scan the result.

Replacing Legacy sem

sem get workflows -p <project>     -> sem-ai workflow list --project <project>
sem get workflow <workflow-id>     -> sem-ai workflow show <workflow-id>
sem get pipelines -p <project>     -> sem-ai pipeline list --project <project>
sem get pipeline <pipeline-id>     -> sem-ai pipeline show <pipeline-id>
sem get jobs                       -> sem-ai job list
sem logs <job-id>                  -> sem-ai job log <job-id>
sem stop job <job-id>              -> sem-ai job stop <job-id>
sem stop pipeline <pipeline-id>    -> sem-ai pipeline stop <pipeline-id>
sem rebuild pipeline <pipeline-id> -> sem-ai pipeline rebuild <pipeline-id>
sem rebuild workflow <workflow-id> -> sem-ai workflow rerun <workflow-id>

Keep legacy sem for interactive shell workflows that sem-ai does not replace yet:

sem attach <job-id>
sem debug <job-id>

Safety

• Inspect first, mutate second.
• Promotion commands require --confirm; verify --target, --param, and the pipeline id before execution.
• Prefer sem-ai <command> --help over guessing flags.
• Use sem-ai yaml validate before pushing pipeline config changes.

Reference: https://docs.semaphore.io/reference/sem-ai-cli