profile

name	profile
description	Load a selective config profile by role (engineer/business/content/ops)
when_to_use	Use when loading a selective config profile by role — engineer, business, content, or ops — trimming the active agent/MCP surface to what that role needs. Triggers on: profile <role>, load the business profile, switch to content mode.
argument-hint	<role>

Profile — Selective Config Loading

Manage configuration profiles that control which agents, skills, and rules are active.

Usage

• /profile — show current profile and available profiles
• /profile core — switch to core profile
• /profile security — switch to security profile
• /profile ali — switch to Ali's personal profile
• /profile create <name> — create a new profile
• /profile show <name> — show what's in a profile

Argument: $ARGUMENTS

Instructions

Parse arguments:

• No args → show current profile + list available profiles
• Profile name → activate that profile
• create <name> → create new profile interactively
• show <name> → display profile contents

What is a Profile?

A profile is a named subset of the Kun engine configuration. Instead of loading all 44 agents, 40+ skills, and all rules, a profile selects only what's relevant for a specific role, task, or person.

Profiles are stored in ~/.claude/profiles/ as YAML files.

Built-in Profiles

core — Minimum viable engine

The essentials for any developer on any repo.

# ~/.claude/profiles/core.yaml
name: core
description: Essential tools for daily development
agents:
  - nextjs
  - react
  - typescript
  - build
  - deploy
  - git
  - github
commands:
  - dev
  - build
  - deploy
  - clone
rules: [] # all rules always apply

developer — Full development toolkit

Everything a developer needs, minus business agents.

name: developer
description: Full development toolkit
agents:
  - nextjs
  - react
  - typescript
  - tailwind
  - prisma
  - shadcn
  - authjs
  - build
  - deploy
  - test
  - git
  - github
  - architecture
  - pattern
  - structure
  - middleware
  - internationalization
  - semantic
  - sse
  - performance
  - atom
  - template
  - block
commands:
  - dev
  - build
  - deploy
  - test
  - clone
  - atom
  - block
  - template
  - schema
  - wire
  - code
  - check
  - translate
  - docs
  - security
  - performance
rules: all

security — Security and compliance focus

For security audits and auth work.

name: security
description: Security audit and auth focus
agents:
  - guardian
  - authjs
  - middleware
  - typescript
  - build
  - test
commands:
  - security
  - build
  - test
  - check
rules: all

business — Non-technical operations

For Ali, Samia, Sedon — business operations via Cowork.

name: business
description: Business operations — no code agents
agents:
  - captain
  - revenue
  - growth
  - support
  - product
  - analyst
commands:
  - weekly
  - monitor
  - costs
  - pricing
  - proposal
  - content-calendar
  - dispatch
  - issue
rules: []

qa — Quality assurance

For Ali's testing workflow.

name: qa
description: QA and testing workflow
agents:
  - quality
  - report
  - test
  - guardian
  - build
  - sse
commands:
  - test
  - build
  - check
  - security
  - performance
  - screenshot
rules: all

full — Everything

All agents, all skills, all rules. Default for Abdout.

name: full
description: Complete engine — all agents, skills, rules
agents: all
commands: all
rules: all

Personal Profiles

abdout — Builder (alias for full)

name: abdout
extends: full

ali — QA + Sales

name: ali
extends: qa
agents_add:
  - revenue
  - support
  - analyst
commands_add:
  - report
  - monitor
  - proposal

samia — R&D + Kun Care

name: samia
extends: business
agents_add:
  - tech-lead
  - learn
  - analyze
commands_add:
  - learn
  - analyze
  - docs

sedon — Executor

name: sedon
extends: business
agents_add:
  - ops
  - deploy
commands_add:
  - deploy
  - monitor
  - incident
  - credentials

Creating a Profile

/profile create <name>:

1. Ask: "What's the role? (developer, business, security, custom)"
2. Ask: "Which agents? (list, or 'like ')"
3. Ask: "Which commands?"
4. Generate YAML file in ~/.claude/profiles/<name>.yaml
5. Report: "Profile created with N agents, M commands."

Showing a Profile

/profile show <name>:

Display the profile contents in a readable table:

Profile: ali (QA + Sales)
Extends: qa

Agents (12):
  quality, report, test, guardian, build, sse,
  revenue, support, analyst

Commands (10):
  test, build, check, security, performance, screenshot,
  report, monitor, proposal

Rules: all

Switching Profiles

/profile <name>:

1. Read the profile YAML
2. Report what's being activated/deactivated
3. Note: Profile switching is informational for now — it tells agents which subset to use. Full enforcement requires settings.json allowedTools support (Phase 2).

Profile in Analyze

When /analyze generates config for a repo, it includes profile recommendations:

## Recommended Profiles

- **core**: nextjs, typescript, build, deploy — for quick fixes
- **full**: + prisma, tailwind, shadcn, test, i18n — for feature work

How Profiles Work (Current)

Profiles are advisory in Phase 1:

• The YAML files define what should be loaded
• Agents read the active profile and scope their work accordingly
• The captain uses profiles for team allocation

Profiles become enforced in Phase 2:

• Settings.json allowedTools maps to profile
• Only agents in the active profile get spawned
• Hook validates agent spawning against profile

Rule: Not everyone needs 44 agents. Profiles give each person and each task exactly the engine they need — no more, no less.