CLIProxyAPI is a powerful proxy server tool that provides a unified API interface for various AI services (Gemini, Claude, OpenAI, etc.). This article focuses on how to configure a Google Pro Account via CLIProxyAPI, using OAuth authentication to seamlessly invoke Gemini’s advanced models.

1. Preparation

Before starting, ensure you have compiled or downloaded the cli-proxy-api executable file.

2. Prepare Configuration File

First, we need to prepare the config.yaml configuration file.

X 2026 Core Algorithm Rules - Quick Cheat Sheet

Update Date: January 24, 2026 Source: X Official Open Source Algorithm, Industry Analysis Consolidation

🎯 X 2026 Algorithm Summary in 3 Sentences

1. Engagement Velocity in the first 30 minutes determines the life or death of a tweet.
2. Video content gets 2-3x exposure boost, Premium accounts get 4x internal network + 2x external network boost.
3. Non-Premium accounts posting links = Absolute Zero Exposure (Fatal restriction).

📊 The Four Core Factors of the Algorithm

Priority 1️⃣ - Critical (70% Weight)

Priority 2️⃣ - Important (25% Weight)

Priority 3️⃣ - Reference (5% Weight)

Fatal Penalties ⚠️ - Zero Exposure

⚙️ How the Algorithm Works: Workflow

The 3 Stages of a Tweet:

【Stage 1】 Initial Screening (0-30 Mins) - Life or Death Moment
  ↓
  Algorithm selects 1,500 candidate tweets
  50% from accounts you follow
  50% from algorithmic recommendations
  ↓
  Based on: Engagement Velocity, Author Credibility, Content Type
  ↓
  Shown to 100-1000 "Test Users"
  ↓
  If Test User Engagement > 5%: Recommended to more people
  If Test User Engagement < 2%: Tweet sinks

【Stage 2】 Small Traffic (30 Mins - 6 Hours) - Reputation Building
  ↓
  Based on Stage 1 results, tweet shown to broader audience
  More engagement → Algorithm values it more → Recommends to more people

【Stage 3】 Large Traffic (6 Hours+) - Exponential Growth or Sink
  ↓
  Either goes viral (widely shared)
  Or sinks (hidden forever)

💡 Five Key Numbers

✅ Must-Do Checklist (Follow Algorithm for Fast Growth)

Daily Standards

• Every tweet has an image or video (Text-only has lowest exposure)
• Reply to comments within first 30 mins (Build initial momentum)
• Post 15-20 times/day (for Chinese users) or 8-15 times/day (for English users)
• Interact with 30-50 other tweets daily (Participate in community)
• Reply to 80%+ follower comments within 24 hours (Maintain credibility)

Account Preparation

• Upgrade to Premium (Strongly recommended, 4x exposure)
• Complete Profile (Clear bio + Link in Bio)
• Pin top tweet (Showcase best content)
• Avoid links in tweets (Non-Premium = Zero Exposure)

Content Strategy

• 70% Original Insights + 20% Interactive Content + 10% Personal Stories
• At least 1 Thread per week (6-10 tweets connected)
• At least 3-5 Video posts per week
• Participate in daily trending topics (Breaking news highest priority)

🚀 Fast Growth Path

0-1000 Followers (Cold Start, 3 Months)

Key Actions:

During the iOS app submission process for App Store review, developers often encounter various reasons for rejection. Rejections related to Guideline 2.1 typically involve issues with metadata or functional completeness.

Recently, when submitting an app using StoreKit 2 for subscriptions, I encountered a rejection related to PassKit (Apple Pay). This article details the phenomenon, root cause analysis, and the final solution, hoping to help developers facing the same issue.

1. Issue Description

After submitting the app for review, I received the following feedback from the Apple review team:

The App Store review for in-app subscriptions (In-App Purchase, Auto-Renewable Subscriptions) is extremely strict, especially regarding information disclosure. If you received a rejection for Guideline 3.1.2 - Business - Payments - Subscriptions, it’s usually because your subscription page is missing certain necessary legal terms or explanations.

This article details how to fix missing subscription information using a real-world case and ensures compliance with Apple’s review guidelines.

1. Issue Description

Reviewers typically provide rejection reasons similar to the following:

You will learn:

• How to manually install the Claudian plugin into Obsidian
• Configuring the Claude Code executable path
• Methods to use Claude Code within Obsidian
• Troubleshooting common issues
• Best practices and usage tips

What is the Claudian Plugin?

Claudian is an Obsidian community plugin that integrates the Claude Code CLI tool into the Obsidian interface. With Claudian, you can invoke Claude Code directly while editing notes for AI-assisted writing, code generation, content rewriting, and more, without switching between the terminal and Obsidian. The plugin supports context awareness, understanding current note content to provide relevant suggestions.

You will learn:

• How to install the Obsidian Skills plugin
• Methods to generate Canvas via natural language instructions
• Practical Case: Auto-generating a historical knowledge graph
• Canvas file format and custom styling
• More practical scenarios and best practices

What is Obsidian Canvas Automation?

Obsidian Canvas Automation refers to using Claude Code’s Agent capabilities and the obsidian-skills plugin to automatically research materials, organize logical relationships, and generate structured Canvas knowledge graphs via natural language instructions. This workflow combines AI search capabilities, logical reasoning, and Canvas visualization, reducing the time from “idea to knowledge graph” from hours to minutes.

OpenCode is a powerful AI-driven programming assistant that supports code generation, refactoring, debugging, and documentation. This article will guide you from scratch through the installation, configuration, and first run of OpenCode, ensuring you can quickly get started and unlock its full potential.

What is OpenCode?

OpenCode is an open-source AI programming assistant that integrates large language model capabilities deeply with code editors, allowing developers to complete complex programming tasks using natural language.

This article introduces how to integrate the Antigravity proxy service into OpenCode, allowing you to use top-tier large models including Claude Sonnet 4.5 and Gemini 3 Pro for free.

1. Install Plugin

You need to install the opencode-antigravity-auth plugin and configure model definitions.

Method 1: Automatic Installation (Recommended)

Directly enter the following command in the OpenCode conversation to let the Agent automatically execute installation and configuration:

Install the opencode-antigravity-auth plugin and add the Antigravity model definitions to ~/.config/opencode/opencode.json by following: https://raw.githubusercontent.com/NoeFabris/opencode-antigravity-auth/dev/README.md

Method 2: Manual Configuration

If you prefer manual control, you can edit the configuration file:

You will learn:

• What MCP is and how it works
• How to configure local MCP services (e.g., shadcn)
• How to configure remote MCP services (e.g., Context7)
• How to mix multiple MCP service configurations
• Verification and troubleshooting tips

What is MCP (Model Context Protocol)?

Model Context Protocol (MCP) is a standardized protocol enabling AI models to access external contexts and tools. Through MCP, OpenCode can connect to local scripts, remote knowledge bases, or specialized services, thereby understanding domain-specific knowledge or invoking external tools. The protocol supports both local command execution and remote HTTP endpoint access, utilizing authentication via OAuth or API Keys.

You will learn:

• What OpenCode Skills are and their function
• How to create a standard Skills directory structure
• How to import and verify skill packages
• Troubleshooting common issues
• How to create custom skill packages

What are OpenCode Skills?

OpenCode Skills are OpenCode’s modular extension mechanism, injecting domain-specific expertise and workflows into the AI assistant through structured instruction packages. Each Skill encapsulates industry best practices, checklists, and execution templates, enabling OpenCode to perform tasks like a subject matter expert. Common Skills include SEO audits, code reviews, test writing, API design, and other professional capabilities.