pixee system-status

PREREQUISITE: Read ../pixee-shared/SKILL.md for global flags, exit codes, and error handling. See ../pixee-auth/SKILL.md if authentication needs to be configured.

pixee system-status reports the health of the connected Pixee deployment itself, distinct from any repository, scan, or finding: platform components, license state, and which AI models are currently available for fix generation. Reach for it when a command that should work is failing in a way that looks like a deployment problem rather than a request problem — an expired license or an unavailable model can produce errors that otherwise look like ordinary API failures.

pixee system-status list

pixee system-status list

No flags beyond the global ones. Pagination is transparent — the CLI walks every page in one call.

Text output is tab-separated with columns type, name, status, description, id. The type discriminator distinguishes platform components (version), license entries (license), and AI model entries (ai-model) in a single flat listing. A healthy entry reports status: "up". Use --output json (or --json) for the full HAL record per entry, which adds type-specific fields not shown in text mode: version on version entries, expires_at and features_enabled on license entries, and model (the underlying model identifier, distinct from the human-readable name) on ai-model entries.

Examples

# Full status listing
pixee system-status list

# Machine-readable, filtered to entries that are not up
pixee system-status list --json | jq '.[] | select(.status != "up")'

# Check whether a specific AI model is currently available
pixee system-status list --json | jq '.[] | select(.type == "ai-model" and .model == "gpt-5.4-mini")'

Best practices

• Check pixee system-status list before escalating an otherwise-unexplained failure across many repositories or scans — a single unhealthy component or expired license can present as scattered per-request errors.
• Filter on type and status in jq rather than assuming a fixed row order; the set of reported components and models can grow between releases.
• To key on the underlying AI model, filter --json output on .model, not .name — name is a human-readable label (e.g. "Analysis Service - Fast Model") and does not contain the model identifier.