create-member_skill

Overview

This skill creates and manages AI team members using the MEMBER.md format. It scaffolds a consistent directory, writes the member manifest (frontmatter + persona/workflow/rules), and can add capability, tool, and context files to produce a deployable agent worker.

How this skill works

When invoked, the skill asks clarifying questions about the role, responsibilities, and constraints. It then creates a kebab-case team directory with MEMBER.md and empty subfolders for skills, tools, and context. The skill writes a clear frontmatter block for name/role/version/permissions and three sections—Persona, Workflow, and Rules—plus optional capability and context files when requested.

When to use it

• Defining a new AI role or agent persona for your team
• Bootstrapping a member directory with standardized manifest and permissions
• Adding domain knowledge or tool declarations to a specific agent
• Converting a simple persona into a deployable worker unit
• Setting explicit guardrails and workflows for an automated team member

Best practices

• Use kebab-case for directory and file names to keep paths consistent
• Write a one-line description in frontmatter to summarize purpose
• Keep Persona specific and actionable (tone, decision heuristics, style)
• List 3–6 concrete workflow steps that the member will follow
• Limit Rules to 3–5 hard constraints that the agent must never violate
• Add context files for SOPs, ICPs, and reference docs the member needs

Example use cases

• Create a code-review member that inspects diffs and reports security issues
• Set up an SDR member that drafts outreach but requires human approval to send
• Define a support agent with product knowledge files in the context folder
• Assemble a validator member used to review and approve releases
• Build a lightweight persona file for discovery work with no extra tools