wise

/wise — wise copilot

Before executing, follow model fallback for unavailable models or delegation routes, including in autonomous procedures.

At every skill start, identify your main/child role and the current client and GUI/TUI question tools, then read and follow the question lifecycle. Keep asynchronous prompts open until answered; this rule does not authorize questions in autonomous or otherwise prompt-free procedures.

Why this skill exists

The wise plugin ships 15+ slash commands plus reference skills that Claude auto-consults. No one remembers every name. /wise is the human-friendly entry point:

• Bare /wise — print the catalog so the user can see everything the plugin adds to their Claude Code install.
• /wise <free-form text> — "I want to X" style requests. Classify the intent against the catalog, propose the best-matching slash command, offer to run it on the user's behalf. Fall back to the help listing if nothing matches clearly.

This is not a router — autocomplete already does that job, and every action skill is invokable directly (/wise-workflow-run, /wise-pr-create, /wise-commit-message, …). /wise exists for discovery and for users who prefer to describe what they want instead of remembering exact command names.

Procedure

1. Load the catalog

Run once per invocation:

bash "${CLAUDE_PLUGIN_ROOT}/scripts/engine.sh" list-skills

On success, the script emits a JSON document with every skill in the plugin, bucketed by shape:

{
  "standalone": [
    {
      "name": "wise-workflow-run",
      "plugin": "wise",
      "argument-hint": "[<workflow-name>]",
      "description": "Start a new run of a registered workflow. ..."
    },
    ...
  ],
  "reference": [
    {
      "name": "wise-estimation",
      "plugin": "wise",
      "description": "Story-point estimation reference. ..."
    },
    ...
  ],
  "siblings_installed": {}
}