
cpa-codex-manager-openai-pool
Manage OpenAI account pools with automated registration, CLIProxyAPI monitoring, and intelligent auto-replenishment
- Installs
- 252
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/reason-machines/codex-skills --skill cpa-codex-manager-openai-pool- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads cpa-codex-manager-openai-pool automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
CPA-Codex-Manager OpenAI Account Pool Management
Skill by ara.so — Codex Skills collection.
CPA-Codex-Manager is a high-performance management panel for OpenAI account pools, featuring automated batch registration, real-time CLIProxyAPI platform monitoring, and intelligent maintenance systems.
What It Does
- Batch Account Registration: Parallel (up to 50 threads, 1000 tasks) and pipeline modes with random intervals
- CLIProxyAPI Auto-Inspection: Detects 401 auth failures and quota exhaustion, auto-removes invalid accounts
- Smart Auto-Replenishment: Monitors pool health, triggers automated replenishment when accounts fall below threshold
- Real-time Monitoring: WebSocket-based log streaming and progress visualization
- Multi-Email Support: Integrates Outlook, TempMail, CloudMail services
- Emergency Defense: Dynamic threshold protection with configurable cooling periods
Installation
Standard Installation
# Using uv (recommended)
uv sync
# Or using pip
pip install -r requirements.txt
Environment Configuration
cp .env.example .env
Edit .env with your configuration:
# Web UI Configuration
WEBUI_HOST=0.0.0.0
WEBUI_PORT=8000
WEBUI_ACCESS_PASSWORD=your_secure_password
# Database (SQLite default)
APP_DATABASE_URL=data/database.db
# Or PostgreSQL
# APP_DATABASE_URL=postgresql://user:password@host:5432/dbname
Docker Installation
mkdir -p ~/CPA-Codex-Manager/data ~/CPA-Codex-Manager/logs
cd ~/CPA-Codex-Manager
curl -O https://raw.githubusercontent.com/Maoleio/CPA-Codex-Manager/main/docker-compose.yml
docker compose up -d
Desktop Mode
pip install pywebview
python desktop.py
Running the Application
Web Mode
python webui.py
Access at http://localhost:8000
Docker Mode
docker compose up -d
docker compose logs -f # View logs
docker compose pull && docker compose up -d # Update
Core Python API Usage
Database Models
from app.models import (
AccountPool,
RegistrationTask,
CPAService,
InspectionHistory
)
from app.database import get_db
# Query account pool
with get_db() as db:
accounts = db.query(AccountPool).filter(
AccountPool.status == 'ready'
).all()
print(f"Ready accounts: {len(accounts)}")
for acc in accounts:
print(f"Email: {acc.email}, Token: {acc.refresh_token}")
Registration Task Creation
from app.services.registration import RegistrationService
from app.schemas import RegistrationTaskCreate
async def create_batch_registration():
service = RegistrationService()
task_data = RegistrationTaskCreate(
count=100,
email_service="outlook", # outlook, tempmail, cloudmail
mode="parallel", # parallel or pipeline
max_workers=50,
random_interval_min=5,
random_interval_max=15
)
task = await service.create_task(task_data)
print(f"Task created: {task.id}")
# Start registration
await service.start_task(task.id)
CPA Service Integration
from app.services.cpa_service import CPAServiceManager
async def configure_cpa():
manager = CPAServiceManager()
# Add CPA service
cpa = await manager.add_service(
name="Production CPA",
base_url="https://your-cpa-instance.com",
admin_token="${CPA_ADMIN_TOKEN}",
enabled=True
)
# Configure auto-inspection
await manager.update_inspection_config(
cpa.id,
enabled=True,
interval_minutes=60,
check_401=True,
check_quota=True,
auto_remove=True
)
Auto-Replenishment Configuration
from app.services.replenishment import ReplenishmentService
async def setup_auto_replenish():
service = ReplenishmentService()
# Configure replenishment rules
await service.configure(
cpa_service_id=1,
enabled=True,
threshold=50, # Trigger when ready accounts < 50
replenish_count=100,
mode="parallel",
email_service="outlook"
)
Manual Account Inspection
from app.services.inspection import InspectionService
async def run_inspection():
service = InspectionService()
result = await service.inspect_cpa_accounts(
cpa_service_id=1,
check_401=True,
check_quota=True,
auto_remove=True
)
print(f"Inspection complete:")
print(f" Total checked: {result['total_checked']}")
print(f" Valid: {result['valid_count']}")
print(f" Invalid (401): {result['invalid_401']}")
print(f" Quota exhausted: {result['quota_exhausted']}")
print(f" Removed: {result['removed_count']}")
Emergency Defense Configuration
from app.services.emergency import EmergencyDefenseService
async def configure_emergency_defense():
service = EmergencyDefenseService()
await service.configure(
enabled=True,
threshold_percentage=50, # Trigger if ready% < 50%
cleanup_percentage=50, # Remove 50% of accounts
cooldown_minutes=5 # Wait 5min before retry
)
WebSocket Integration for Real-time Logs
from fastapi import WebSocket
import asyncio
# Server-side WebSocket handler
async def websocket_log_stream(websocket: WebSocket, task_id: int):
await websocket.accept()
# Stream registration logs
async for log_line in get_task_logs(task_id):
await websocket.send_json({
"type": "log",
"task_id": task_id,
"message": log_line,
"timestamp": datetime.utcnow().isoformat()
})
await websocket.close()
Common Patterns
Full Registration Workflow
from app.services.registration import RegistrationService
from app.services.cpa_service import CPAServiceManager
async def full_registration_workflow():
# Create and start registration task
reg_service = RegistrationService()
task = await reg_service.create_task(
count=200,
email_service="outlook",
mode="parallel",
max_workers=50
)
# Monitor progress
await reg_service.start_task(task.id)
while True:
status = await reg_service.get_task_status(task.id)
print(f"Progress: {status['success']}/{status['total']} "
f"({status['progress_percentage']:.1f}%)")
if status['status'] in ['completed', 'failed']:
break
await asyncio.sleep(5)
# Upload successful accounts to CPA
cpa_manager = CPAServiceManager()
uploaded = await cpa_manager.upload_accounts(
cpa_service_id=1,
task_id=task.id
)
print(f"Uploaded {uploaded} accounts to CPA")
Scheduled Inspection with Auto-Replenish
import asyncio
from app.services.inspection import InspectionService
from app.services.replenishment import ReplenishmentService
async def scheduled_maintenance():
inspection = InspectionService()
replenish = ReplenishmentService()
while True:
# Run inspection
result = await inspection.inspect_cpa_accounts(
cpa_service_id=1,
check_401=True,
check_quota=True,
auto_remove=True
)
# Check if replenishment needed
ready_count = result['valid_count']
if ready_count < 50:
print(f"Low account count ({ready_count}), triggering replenishment")
await replenish.trigger_replenish(
cpa_service_id=1,
count=100
)
# Wait 1 hour
await asyncio.sleep(3600)
Query Account Pool Statistics
from app.database import get_db
from app.models import AccountPool
from sqlalchemy import func
def get_pool_statistics():
with get_db() as db:
stats = db.query(
AccountPool.status,
func.count(AccountPool.id).label('count')
).group_by(AccountPool.status).all()
return {status: count for status, count in stats}
# Example output: {'ready': 150, 'invalid': 10, 'quota_exhausted': 5}
Configuration Best Practices
Inspection Configuration
# config.py or .env
INSPECTION_CONFIG = {
"interval_minutes": 60, # Check every hour
"check_401": True, # Enable auth failure detection
"check_quota": True, # Enable quota exhaustion detection
"auto_remove": True, # Auto-remove invalid accounts
"emergency_threshold": 50, # Trigger emergency if ready% < 50%
}
Replenishment Configuration
REPLENISHMENT_CONFIG = {
"enabled": True,
"threshold": 50, # Trigger when ready accounts < 50
"replenish_count": 100, # Add 100 accounts per trigger
"mode": "parallel", # Use parallel mode for speed
"email_service": "outlook",
"max_workers": 50
}
Registration Modes
Parallel Mode (fast, high volume):
parallel_config = {
"mode": "parallel",
"max_workers": 50,
"count": 1000
}
Pipeline Mode (stealth, bypass rate limits):
pipeline_config = {
"mode": "pipeline",
"random_interval_min": 10,
"random_interval_max": 30,
"count": 100
}
Troubleshooting
Registration Failures
# Check failed registrations
from app.models import RegistrationTask
with get_db() as db:
failed_tasks = db.query(RegistrationTask).filter(
RegistrationTask.status == 'failed'
).all()
for task in failed_tasks:
print(f"Task {task.id}: {task.error_message}")
CPA Connection Issues
from app.services.cpa_service import CPAServiceManager
async def test_cpa_connection():
manager = CPAServiceManager()
try:
health = await manager.check_health(cpa_service_id=1)
print(f"CPA Status: {health['status']}")
except Exception as e:
print(f"Connection failed: {e}")
# Check base_url and admin_token in database
Account Pool Health Check
from app.services.inspection import InspectionService
async def health_check():
service = InspectionService()
result = await service.inspect_cpa_accounts(
cpa_service_id=1,
check_401=True,
check_quota=True,
auto_remove=False # Don't remove, just check
)
total = result['total_checked']
valid = result['valid_count']
health_percentage = (valid / total * 100) if total > 0 else 0
print(f"Pool health: {health_percentage:.1f}%")
if health_percentage < 50:
print("⚠️ Pool health critical, consider manual intervention")
Database Migration
# If using PostgreSQL, migrate from SQLite
from app.database import init_db
from app.models import Base
# Initialize new database schema
init_db()
# Manual data migration (example)
import sqlite3
from app.database import get_db
sqlite_conn = sqlite3.connect('data/database.db')
sqlite_cursor = sqlite_conn.cursor()
accounts = sqlite_cursor.execute(
"SELECT email, password, refresh_token, status FROM account_pool"
).fetchall()
with get_db() as db:
for email, password, token, status in accounts:
account = AccountPool(
email=email,
password=password,
refresh_token=token,
status=status
)
db.add(account)
db.commit()
Desktop Packaging
macOS
chmod +x scripts/build_macos_dmg.sh
./scripts/build_macos_dmg.sh
# Output: dist/CPA-Codex-Manager.dmg
Windows
scripts\build_windows.bat
REM Output: dist\CPA-Codex-Manager\CPA-Codex-Manager.exe
Environment Variables Reference
| Variable | Default | Description |
|---|---|---|
WEBUI_HOST | 127.0.0.1 | Web UI bind address |
WEBUI_PORT | 8000 | Web UI port |
WEBUI_ACCESS_PASSWORD | - | Admin password (required) |
APP_DATABASE_URL | data/database.db | Database connection string |
All secrets should be stored in environment variables or .env file, never hardcoded.
FAQ
What does the cpa-codex-manager-openai-pool skill do?
Manage OpenAI account pools with automated registration, CLIProxyAPI monitoring, and intelligent auto-replenishment
How do I install cpa-codex-manager-openai-pool?
Run npx skills add https://github.com/reason-machines/codex-skills --skill cpa-codex-manager-openai-pool in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/cpa-codex-manager-openai-pool/ (personal) or .claude/skills/cpa-codex-manager-openai-pool/ (project).
Which AI agents can use cpa-codex-manager-openai-pool?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is cpa-codex-manager-openai-pool safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains cpa-codex-manager-openai-pool?
It is published by reason-machines in reason-machines/codex-skills. It has 252 recorded installs.
More from reason-machines/codex-skills
academic-research-skills-codex
reason-machines/codex-skills
AI-assisted academic research workflows for literature review, paper writing, peer review, and research pipelines
codexplusplus-launcher
reason-machines/codex-skills
Enhanced launcher for Codex App that unlocks plugins, enables session deletion, exports conversations, and syncs provider sessions via CDP injection
codex-plugin-claude-code
reason-machines/codex-skills
Use OpenAI Codex from inside Claude Code for code reviews, adversarial reviews, and delegating tasks to Codex as a subagent.
codex-plusplus-tweak-system
reason-machines/codex-skills
Expert in developing, installing, and managing tweaks for the Codex++ desktop app extension system
awesome-codex-skills-curator
reason-machines/codex-skills
Curate, install, and manage Codex skills from the awesome-codex-skills collection for AI coding agents.
awesome-codex-subagents
reason-machines/codex-skills
A curated collection of 136+ specialized Codex subagents for development tasks across 10 categories
Popular ai & agents skills
google-agents-cli-publish
google/agents-cli
This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).
google-agents-cli-eval
google/agents-cli
This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).
skill-creator
anthropics/skills
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
codex-pet
prime-skills/runcomfy-agent-skills
Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.
full-output-enforcement
leonxlnx/taste-skill
Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.
executing-plans
obra/superpowers
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available