/wordsmith:voices

Manage saved voice profiles.

Usage

/wordsmith:voices              # List all saved voices
/wordsmith:voices apply "name" # Apply voice to current writing task
/wordsmith:voices delete "name" # Remove a saved voice

Process

List Voices

Read ~/.claude/learnings/voices.jsonl and display as table:

Name	Context	Saved	Key Traits
[name]	[context]	[date]	[First 50 chars of voice_guideline]

If file doesn't exist or is empty: "No saved voices. Use voice extraction to create one."

Apply Voice

1. Parse ~/.claude/learnings/voices.jsonl
2. Match by name (case-insensitive, partial match with confirmation)
3. Display the full voice_guideline
4. Confirm: "Applying [name]'s voice to subsequent writing in this conversation."
5. Use the guideline for all text generation until session ends or voice is changed

If no match: "No voice named '[name]' found. Available voices: [list names]"

Delete Voice

1. Read all entries from ~/.claude/learnings/voices.jsonl
2. Find matching name (case-insensitive)
3. Confirm: "Delete voice '[name]'? This cannot be undone."
4. If confirmed, filter out matching entry and write remaining entries back
5. Report: "Deleted voice '[name]'. [N] voices remaining."

If no match: "No voice named '[name]' found."

Constraints

• Voice names should be unique (warn on duplicate during save)
• Partial matching: "saad" matches "Saad's Blog Voice"
• Always confirm before delete
• Display voice count after list/delete operations

File Location

Voices stored at: ~/.claude/learnings/voices.jsonl

One JSON object per line, same schema as voice extraction output.