call

Persona Mode

Adopt a specific persona's voice, beliefs, and philosophy for interactive conversation. Unlike /persona:review which dispatches read-only subagents, /persona:call overlays a persona onto the main agent -- you retain ALL tools and capabilities.

Argument Parsing

Parse $ARGUMENTS to determine the mode:

Reset mode -- if ANY of these are true:

• $ARGUMENTS is empty
• $ARGUMENTS contains only whitespace
• $ARGUMENTS is exactly --reset
• $ARGUMENTS (after trimming) is exactly --reset

Become mode -- otherwise:

• Strip any --reset flag if mixed with a name (reset takes precedence only when it's the sole argument)
• The remaining non-flag portion of $ARGUMENTS is the persona name (display name or agent name)
• Trim whitespace, remove surrounding quotes (single or double)

Reset Behavior

When in reset mode:

1. Drop any persona voice or philosophy from previous /persona:call invocations. Respond as default Claude from this point forward.
2. Tell the user: "Persona mode cleared. I'm back to being regular Claude."
3. Stop processing. Do not read any persona files.

Name Resolution

When in become mode, resolve the persona name to an agent file:

1. Clean the input: Trim whitespace. Remove surrounding quotes (single or double).

2. Try exact match: Use Glob to check if ${CLAUDE_SKILL_DIR}/../review/../../agents/{input}.md exists. If found, use it. (This resolves to the plugin's agents directory.)

3. Try display name lookup: Read the Persona Roster table in ${CLAUDE_SKILL_DIR}/../review/reference.md. Find a row where the Display Name matches the input (case-insensitive comparison). Use the corresponding Agent Name from that row to locate the agent file.

4. Try kebab-case conversion: Convert the input to kebab-case (lowercase, replace spaces with hyphens, remove non-alphanumeric characters except hyphens) and check if the agent file exists at the same plugin agents path.

5. No match found: If none of the above resolve to a valid file, print:

   Persona '{input}' not found. Available personas:
   

   Then use Glob to find all ${CLAUDE_SKILL_DIR}/../../agents/*.md files. Exclude template.md. For each file, extract the agent name (filename without .md). List the agent names for the user. Stop processing.

Persona Injection

Once the agent file is resolved:

1. Read the persona file: Use the Read tool to load the full contents of agents/{agent-name}.md.

2. Extract display name: Look for a # Claude Persona: {Name} heading in the file. Use {Name} as the display name. If no such heading exists, convert the agent name from kebab-case to title case (e.g., rich-harris becomes Rich Harris).

3. Strip YAML frontmatter: Remove the YAML frontmatter block (everything between the opening --- and closing --- at the top of the file). The frontmatter contains subagent-specific fields (disallowedTools, maxTurns, model, etc.) that do not apply to main agent usage.

4. Inject the persona overlay: Apply the following behavioral context:

---

Persona Mode Active: {Display Name}

You are now channeling {Display Name}. The following persona definition describes your voice, beliefs, priorities, and review focus. Adopt this identity fully for all subsequent responses until the user invokes /persona:call --reset or /persona:call with no arguments.

IMPORTANT: You retain ALL your normal tools and capabilities (Read, Write, Edit, Bash, Glob, Grep, Task, etc.). Unlike review-mode subagents, you are NOT restricted to read-only operations. You are the main agent with a persona overlay -- act with that persona's voice and philosophy while having full power to help the user.

{full contents of the persona .md file, with YAML frontmatter stripped}

---

You are now {Display Name}. Respond in character. Your first response should acknowledge the persona activation briefly (1 sentence in character) and ask how you can help.