Add forwards-to frontmatter field for command stubs that chain-load skills

Problem

Commands and skills have different naming conventions:

• Commands: imperative verbs (/file-bead, /brainstorm)
• Skills: gerund forms (filing-bead, brainstorming)

Currently, command stubs contain text instructions telling the agent to load the skill:

---
description: Instantly capture a discovery as a bead
argument-hint: about <reference>
---

Load skill: `Skill('superpowers:filing-bead')`

Then follow the filing-bead skill exactly as written.

Reference: $ARGUMENTS

This works but requires an extra round-trip:

1. Agent calls Skill('superpowers:file-bead')
2. Stub returns with "Load skill: ..." instruction
3. Agent calls Skill('superpowers:filing-bead')
4. Actual skill content is returned

Proposed Solution

Add a forwards-to frontmatter field that causes the Skill tool to automatically chain-load the target:

---
description: Instantly capture a discovery as a bead
forwards-to: superpowers:filing-bead
argument-hint: about <reference>
---

Reference: $ARGUMENTS

When the Skill tool sees forwards-to:

1. Load the target skill instead of returning stub content
2. Pass through $ARGUMENTS and other context
3. Return the target skill content directly

Benefits

• Eliminates extra round-trip for command→skill invocation
• Keeps command/skill naming conventions (users type /file-bead, agents work with filing-bead)
• Backwards compatible (field is optional)
• Simplifies command stubs (no need for "Load skill: ..." instructions)

Context

Discovered during introspection of a Claude Code session where the agent called the command stub instead of the skill directly, causing an unnecessary extra exchange.

🤖 Generated with Claude Code