CLAUDE CODE PLUGIN // V.1.1.0 // SHELL-NATIVE

ONE ORCHESTRATOR.
MANY VIRTUOSOS.
ZERO CHAOS.

A production-ready Claude Code plugin. Five specialised AI agents. Model routing by task. Quality gates at every stage. Loop detection. Destructive command blocking. No Node, no Python.

System Status
Agent Roles 5 + 2 ACTIVE
Quality Hooks 6 ENABLED
Model Tiers OPUS / SONNET / HAIKU
Commands /kleiber · /kleiber-brand
Skills 2 BUNDLED
Runtime Deps ZERO
Loop Guard ARMED
$ /kleiber Build user auth

WHY KLEIBER. THE OPINIONATED LAYER CLAUDE CODE LEFT OUT.

Claude Code shipped native agent teams.
Kleiber makes them shippable.
01

Preconfigured Roles

Five battle-tested prompts. Architect, Frontend Engineer, Backend Engineer, Validator, Scribe. Stop rewriting role setup every run.

Spawn time: <3s
02

Model Routing

Opus for architecture. Sonnet for implementation. Haiku for validation. Cheapest reliable tier for every step. Token spend drops without quality loss.

Typical saving: 40-60%
03

Ralph Loop Guard

When an agent makes the same mistake three times in a row, it gets pulled. No infinite token-burning loops at 3 AM.

"I bent my wookie."
04

Quality Hooks

Tests, type checks, lint, and brand-drift checks at every stage. A teammate cannot mark a task complete if validation fails.

6 hooks · stdin JSON API
05

Destructive Block

rm -rf, git push --force, DROP TABLE. Pre-execution. Not after the bloodbath.

Pre-hook · exits 2
06

Delegate Mode

Lead agent coordinates only. Never implements. Scope boundaries stay crisp; no role drift; no "let me just fix this real quick".

Enforced at spawn
Agent Roles

FIVE ROLES. FIXED RESPONSIBILITIES. CLEAR HANDOFFS.

Role Model Writes Code Purpose
Architect OPUS NEVER Design, review, approve. Coordinates the team in delegate mode.
Engineer-Frontend SONNET YES UI components, state, accessibility, tests.
Engineer-Backend SONNET YES Routes, queries, business logic, integration tests.
Validator HAIKU READ-ONLY Runs tests, type checks, linting.
Scribe HAIKU DOCS ONLY Docs, changelog, decision records.
Brand-Analyst SONNET DOCS ONLY Probes ChatGPT / Gemini / Perplexity. Scores brand visibility across AI models.
Brand-Architect OPUS DOCS ONLY Generates optimisation blueprints. Read-only on source; proposals only — humans approve.
Quality Hooks

SIX HOOKS. STDIN JSON. EXIT 2 BLOCKS.

stop-loop-guard Stop Detects and halts Ralph Loops on repeated errors.
task-completed TaskCompleted Blocks completion when tests or type checks fail.
teammate-idle TeammateIdle Prevents idle status when uncommitted changes exist.
post-edit-verify PostToolUse Runs type checks after each Write/Edit.
block-destructive PreToolUse Stops destructive Bash commands before execution.
brand-drift-check PreToolUse Blocks edits that remove the Brand Identity block from CLAUDE.md.
FEATURE SPOTLIGHT

RALPH LOOP GUARD

Agents repeat themselves. When the same error fires three times in a row, the loop guard pulls the teammate, posts a structured report, and routes the work to a different role. Token burn stops at try #4 instead of try #400.

"I bent my wookie."
— Loop Guard, observing an agent on attempt 3
$ /kleiber Build login flow
[ARCHITECT] Spawning team...
[ENG-FE] Building Login.tsx...
[VALIDATOR] Test failed: type 'string' not assignable
[ENG-FE] Retrying...
[VALIDATOR] Test failed: same error
[ENG-FE] Retrying...
[VALIDATOR] Test failed: same error
[LOOP GUARD] ⚠ Ralph Loop detected. Pulling ENG-FE.
[LOOP GUARD] Routing to ARCHITECT for review.
> Estimated tokens saved: ~84,000
$ /kleiber-brand
[BRAND-ANALYST] Probing ChatGPT... ✓
[BRAND-ANALYST] Probing Gemini... ✓
[BRAND-ANALYST] Probing Perplexity... ✓
> Visibility score: 12 / 100
> Recommendations queued: 7
[BRAND-ARCHITECT] Generating blueprint...
[BRAND-ARCHITECT] Writing to docs/blueprint.md
✓ Blueprint ready for human review.
FEATURE SPOTLIGHT

AI VISIBILITY AUDITING

Two extra agents — brand-analyst and brand-architect — probe ChatGPT, Gemini and Perplexity to score how your product shows up in AI answers, then write an optimisation blueprint to docs/.

Source code is read-only to these agents. Every proposal is a draft humans approve. The brand-drift-check pre-hook blocks any edit that strips the Brand Identity block from CLAUDE.md.

/kleiber-brand Read-only on source Drift guard armed
Need the full platform — multi-brand tracking, daily scoring, recommendation share dashboards? → BrandMind.app
Bundled Skills

EVERYTHING PRELOADED. NOTHING TO STITCH.

skills/orchestration
Team coordination playbook

Model routing rules, task decomposition, quality gates, coordination patterns, and anti-pattern detection — everything the Architect uses to keep the team aligned.

skills/brand-visibility
AI search ranking auditor

The methodology behind /kleiber-brand — probe templates, scoring rubric, optimisation playbook. BrandMind-style auditing, built in.

Install

PREREQUISITE: ENABLE EXPERIMENTAL AGENT TEAMS.

// In .claude/settings.json
{
  "env": {
    "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
  }
}
// Or in your shell
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
# Add the marketplace
/plugin marketplace add Devgapperk/kleiber
# Install the plugin
/plugin install kleiber@kleiber-marketplace
Usage

START CLAUDE CODE. RUN THE COMMAND.

// Build features with the full team
/kleiber Build user authentication with login, signup, and password reset
// Audit AI visibility for your product
/kleiber-brand
Example

Implement a payment processing module with Stripe integration.

Example

Review PR #142 with security, performance, and test-coverage teammates.

Example

Investigate a production bug with competing hypotheses and cross-challenge.

Model Cost Reference

CHEAPEST RELIABLE TIER. EVERY TIME.

Opus 4.6
$5 / $25
Per 1M tokens · in / out

Architecture, security, compliance.

Sonnet 4.5
$3 / $15
Per 1M tokens · in / out

Feature implementation, default tier.

Haiku 4.5
$1 / $5
Per 1M tokens · in / out

Validation, docs, boilerplate.

Known Limitations

WHAT KLEIBER WILL NOT DO YET.

  • No session resumption: teammates are not restored by /resume or /rewind.
  • One team per session: teammates cannot spawn additional teams.
  • Task status may lag and require lead-agent nudges.
  • Split-pane requires tmux or iTerm2. In-process mode works everywhere.
Install

STOP REWRITING AGENT SETUP.
SHIP THE FEATURE.