Skills
All management of agento-patronum happens via skills — the open specification for agent capabilities. Skills are invoked as slash commands inside Claude Code.
| Skill | Description |
|---|---|
/patronum-add | Add a pattern to the protection list |
/patronum-remove | Remove a pattern from the protection list |
/patronum-list | Show all protected patterns |
/patronum-suggest | Get stack-specific protection suggestions |
/patronum-verify | Run self-test to verify hook enforcement |
How they work
Each skill is a SKILL.md file following the agentskills.io specification. Skills instruct Claude to call the corresponding Node.js script via the Bash tool. The scripts read and write ~/.claude/patronum/patronum.json.
No external binary, no API calls — everything runs locally via Node.js.