DOCS · Install (CLI)

saferskills list

Show your full local capability inventory across detected agents, annotated with trust scores.

saferskills list shows your full local inventory — every capability discovered across your detected agents, regardless of how it was installed — each annotated with its security score where known. It uses the same discovery the no-target capability audit performs, so it sees capabilities you added by hand, not just ones the CLI installed. On a terminal it then offers to scan the unscanned ones inline.

How do I list installed capabilities?

Run the command with no arguments:

npx saferskills list

The CLI scans each detected agent’s own config directories, collects every capability it finds, and renders one annotated row per capability. The eight supported agents and their config locations are covered in the per-agent guides.

What does each row show?

Each capability is annotated with its security score, and the source of that score depends on how the capability is tracked:

On a TTY, list then offers to scan the unscanned capabilities inline and re-renders with the new scores. In headless mode — --json, --quiet, or --non-interactive — it prints a hint pointing at the capability audit instead of launching the interactive scan.

What flags and exit codes apply?

The global flags apply. --json emits the inventory as machine-readable data on stdout and suppresses the inline-scan prompt; --quiet and --non-interactive do the same. Exit codes follow the shared convention — 0 on success, 4 if an agent config directory can’t be read, 6 if a score lookup needs the API and it is unreachable.

Where do I go next?