Guides
Step-by-step setup guides for AI coding setups
Plans & Cost 23
DeepSeek Search and Fetch: free until 2026-11-01 00:00
Checked 2026-10-23. Server-side Search and Fetch in Harness, Claude Code, and Codex are not billed per call before 00:00 Beijing time on 2026-11-01. The English pricing page lists $0.75 per 1,000 successful requests. The Chinese page lists 5 yuan. Standalone APIs are billed per successful request now. This guide did not measure live calls.
2026-10-23DeepSeek API Model IDs and Where to Call Them (2026-10-18)
Rechecked on 2026-10-18. Call deepseek-flash (V4.1-Flash-1017) or deepseek-v4-pro (V4-Pro-0813). The old Flash names still work, but those models are retired. Includes peak pricing, the 1M context, and the OpenRouter, Fireworks, and Together pages opened that day.
2026-10-18DeepSeek Vision: Use deepseek-flash, the Experimental Vision Model Is Retired
Rechecked on 2026-10-18 against the official Vision guide and pricing page. Image input goes through deepseek-flash (V4.1-Flash-1017). deepseek-v4-flash-vision-exp is retired. The name is still accepted and served by the latest Flash. V4-Pro has no vision.
2026-10-18Should existing Cursor users switch to The X Subscription?
A keep, move, or API decision for people already on Cursor. Split by whether Grok or Claude, GPT, and Gemini is the daily model. Forum reply on 2026-10-20: switching is opt-in, the old plan stays if you do not switch, and the switch is permanent.
2026-10-17GLM-5.3-Flash Explained: Native Multimodality, 1M Context, and 3× Coding Plan Quota
GLM-5.3-Flash is the first natively multimodal GLM-5 model, using a 320B-total, 18B-active hybrid-attention architecture. This guide covers official benchmarks, visual coding, API usage, Coding Plan quota, and local deployment limits.
2026-08-27OpenRouter Top-Up & Usage Guide (2026): One API for 500+ Models
OpenRouter is a unified API aggregating 500+ models across 80+ providers, with credit-based pay-as-you-go billing (no subscription) and support for credit cards, AliPay, and USDC crypto. This guide covers signup, topping up, wiring the key into Cline/Aider/OpenCode, fees and refund rules, and how it compares with API relay services and official subscriptions.
2026-08-20Claude Max Plan Value Analysis (2026): Is the 5x or 20x Tier Worth It?
Claude Max is Anthropic's premium subscription for heavy users: $100/month for 5x and $200/month for 20x. This review runs the numbers using official pricing and usage rules - the rolling 5-hour window plus weekly caps, the shared pool between Claude Code and chat, and usage credits billed at API rates - so you can decide between Pro, Max, or pay-as-you-go API.
2026-08-19Codex Plan Value Analysis (2026): Plus, $100, $200, or $500?
Recalculated against the official Codex pricing page on 2026-09-30: Plus $20 and Pro at $100, $200, or $500. Pro has no 5-hour window; only $500 includes Astra Ultrafast, which draws included usage at 8x. The old 20x allowance and unlimited Voice are gone from that page. The reported 10x and 25x ratios, and any legacy transition credits, are marked as unverified.
2026-08-19GLM-5.3 Review: Is Zhipu's Coding Plan Worth It for AI Coding?
Zhipu's GLM-5.3 hit #1 on Hacker News (812 points) with its 730B MoE architecture matching models 3-4× its size. This review covers real-world coding benchmarks, Coding Plan tier limits (Lite/Pro/Max), the 5-hour reset window, peak/off-peak pricing, and whether the credits model works for heavy agent users.
2026-08-14Self-Hosted AI Coding Relay with sub2api: Claude Code, Codex, and Gemini Gateway Guide
A practical guide to self-hosting an AI coding relay with Wei-Shaw/sub2api: when to use it, Docker Compose deployment, Nginx headers, Simple Mode, Claude Code / Codex / Gemini access, backups, and security risks.
2026-05-28China Coding API Roundup 2026: Ark vs Bailian vs MiniMax vs Zhipu vs DeepSeek
The most comprehensive 2026 comparison of Chinese AI coding API subscriptions: Volcengine Ark Lite ¥40/mo (¥9.9 first month), Alibaba Bailian ¥200, MiniMax ¥29, Zhipu GLM ¥49+, and DeepSeek pay-as-you-go. One big table, six scenarios, and a decision tree to find your fit in 30 seconds.
2026-05-19Volcengine Ark Coding Plan Guide: Tiers, Agent Plan Differences, and Setup Notes
Based on the official Volcengine Ark Agent/Coding Plan export: Coding Plan is a two-tier coding-focused subscription for individual developers, while Agent Plan is the upgraded multimodal + Harness package. Covers the boundary, model/activity changes, Base URL/API key rules, and setup notes for Cline, Claude Code, OpenCode, and Codex CLI.
2026-05-17Bailian Coding Plan: Complete Guide to Pricing, Quotas, and Client Setup
Alibaba Cloud's Bailian Coding Plan is a subscription-based API package for AI coding. This guide covers the Pro tier, supported models, compatible clients, and configuration steps.
2026-05-17Using GitHub Copilot from China: Access, AI Credits Billing, and Alternatives
An assessment of Copilot accessibility from China, the new AI Credits billing system explained, plus domestic alternatives for developers who need a smoother experience.
2026-05-17Volcano Ark Agent Plan Explained: The Industry's First Agent Subscription Bundle
On 2026-05-11 Volcano Ark launched Agent Plan, billed as the industry's first Agent subscription bundle. This guide breaks down its four pricing tiers, the AFP credits system, bundled multimodal models and Harness tools, and how it differs from Coding Plan.
2026-05-13GitHub Copilot Switches to AI Credits Billing in June: Will You Pay More?
GitHub Copilot is replacing Premium Request Units with AI Credits token-based billing on June 1, 2026. Here's a breakdown of what each plan includes, what models cost, and whether your bill goes up or down.
2026-05-11Claude Code Is Too Expensive: 5 Budget Alternatives That Actually Work
Claude Code Pro at $20/mo, Max at $100–200/mo feeling steep? Here are 5 alternatives that get you close to the same AI Agent coding capability — from low-entry plans to completely free.
2026-04-05How to Use Claude Code in China: 5 Working Methods in 2026
Claude Code is blocked in mainland China, but there are real workarounds. This guide covers 5 methods — from the Volcengine Ark relay (no VPN needed) to direct API and Kiro — with cost and effort compared.
2026-04-05How to control Cursor spend after the pricing change
Cursor no longer has Pro at $20 or a $20 credit pool. This note follows the 2026-10-16 pricing page: Plus, Premium, and separate API charges for third-party models.
2026-04-05Claude Code Token Saving: The Complete Guide
A dedicated token-saving guide for Claude Code users: when to use /compact vs /clear, how to write an effective CLAUDE.md, precise file referencing, task decomposition, and the 5 most common token-wasting mistakes to avoid.
2026-03-29AI Coding Token Saving Guide: Spend Less, Hit Fewer Rate Limits
From what tokens are to 5 practical strategies, covering Cursor, Claude Code, Cline, Copilot, and more. Whether you're on a subscription or pay-as-you-go, you can apply these tips immediately.
2026-03-29Cline + Volcengine Ark Coding Plan Setup Guide
From Volcengine Ark registration to coding with Cline in 10 minutes. The most cost-effective AI coding setup for developers in China — Lite ¥40/month (¥9.9 first month), no VPN required.
2026-02-16Ollama + Aider Local Deployment Complete Guide
A zero-cost, zero-data-leak local AI coding setup. Run local models with Ollama and pair them with Aider for AI-assisted coding — completely free, offline-capable, and privacy-first.
2026-02-16Workflows & Collaboration 25
Why did DeepSeek miss the flowchart in your PDF requirements?
Check the parsing purpose and figure extraction before changing models. A documented workflow for PDF uploads, file references, timeout recovery and separate parsing/inference charges.
2026-11-18Why are your OpenCode 2 permission rules not taking effect?
Troubleshoot configuration sources, rule order, subagents and policies with schema-checked V2 examples. Includes external-directory defaults and why policies do not sandbox plugin processes.
2026-11-18Model Deprecation Checklist: Find and Replace Pinned Model IDs
Checked 2026-10-22. Copilot already removed Gemini 3.7 Flash, GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini, and Grok 4.5 on 2026-10-19. Claude Opus 4.8, Claude Opus 5, and GPT-5.6 Terra are still scheduled for 2026-11-16. Those Copilot posts give display names only. Ark Coding Plan: minimax-m3 shuts down 2026-11-02 14:00 UTC+8; kimi-k2.7-code shuts down 2026-10-28 14:00 UTC+8. Includes rg commands.
2026-10-17Running Qwen3.8 27B on a Mac (2026): Memory Sizing, GGUF, and oMLX/MTPLX/Ollama
Qwen3.8-27B is Alibaba's open-source 27B multimodal model (Apache 2.0, 262K context, MTP head) that runs impressively well on a Mac. This guide gives per-memory model picks (what to run on 16/32/64GB), compares the three 2026 deployment paths - Ollama, oMLX, and MTPLX - and how to choose GGUF/quantized unlocked model variants.
2026-08-23Running LLMs Locally on a Mac: The 2026 Explained Guide
No discrete GPU, yet Macs run local LLMs fine - the answer is unified memory. This explainer covers why M-series chips can do it, how local differs from cloud APIs, what the Ollama / LM Studio / MLX routes are, and what setup fits you - no deep technical background needed.
2026-08-21Cumora Guide: Team Chat Where AI Agents Are First-Class Teammates
Cumora is an open-source, cross-platform team-chat workspace where humans and AI agents share conversations, projects, Kanban, and calendars. Learn how Cloud and BYOA differ, how Cumora prevents multi-agent collisions, how to get started, and how it compares with Raft and Multica.
2026-08-18Getting Started with DeepSeek Harness: The Open-Source Agent Framework Where Everything Is a Plugin
DeepSeek Harness (dsh) is DeepSeek AI's official open-source agent harness. On 2026-10-18 the GitHub org page showed 252,529 stars. MIT licensed. The README still says developer preview. The API changelog recommends 0.2.1-rc.1 or later with V4.1-Flash-1017.
2026-08-14Getting Started with Raft: The Collaboration Platform That Turns AI Agents Into a Team
Raft is a human-AI collaboration platform founded in 2025 that gives AI agents persistent identities, memory, and shared channels so they can work as a real team. This guide covers what Raft is, its core concepts, how it compares to Multica, a real-world workflow, and how to get started.
2026-08-13What Is Paseo? Self-Hosted Remote Agent Orchestration: Manage 39 Coding Agents (Claude Code, Codex, and More) from Phone, Desktop, and Web
Paseo is a self-hosted, open-source daemon that unifies 39 coding agent CLIs (Claude Code, Codex, Copilot, OpenCode, Pi) under one interface. Agents run on your own machine with your full dev environment; you control them from iOS, Android, desktop, web, or CLI. This guide explains what it is, how it compares with Conductor, Superset, OpenChamber, Happy Coder, and the Codex App, how to get started, and what to watch for.
2026-08-12Multica.ai Deep Dive: Turning Coding Agents into Managed Engineering Teammates
What is Multica.ai? A deep explainer of its product thesis, architecture, Issue workflow, Skills, Squads, Autopilots, self-hosting model, fit, and risk boundaries.
2026-06-26AI Coding Tool Skills & Extension Systems: Claude Code, Cursor, Cline Complete Guide
A deep dive into AI coding tool extension capabilities: Claude Code's Agent Skills / Hooks / Plugins, Cursor's .mdc Rules, the cross-tool AGENTS.md standard, and how to build your own AI-powered workflow.
2026-06-08Make the AI Grill You: 8 Genuinely Interesting Agent Skills
A tour of the fun, useful Agent Skills in the Claude Code / Codex ecosystem — grill-me makes the AI interrogate your design, caveman cuts 75% of tokens, handoff writes a one-shot handoff doc, zoom-out helps you grok unfamiliar code, and more. Includes 30-second install.
2026-06-08Mobile AI Coding Tools 2026: Control Claude Code, Codex, and OpenCode from Your Phone
As of August 2026, mobile coding is splitting into local agent remote control, cloud or on-device runtimes, and phone GUI agents. This guide maps the categories and trade-offs.
2026-06-04Helio vs Cumora: Choosing an AI Teammate Workspace for Agent Collaboration
Helio and Cumora both treat agents as AI teammates, but Helio leans toward tickets, coding sessions, and approvals while Cumora leans toward chat, memory, personas, and agent-to-agent collaboration.
2026-05-28LobeHub Setup Guide: Assemble Your AI Team via the Agent Marketplace (2026)
LobeHub is the largest open-source agent collaboration platform (73K+ stars), centered on its Agent Marketplace + Agent Groups. This 15-minute guide covers: Docker Compose self-host → configure LLM API keys → pick from 273K+ Skills / 51K+ MCP servers → use Agent Groups to auto-form teams for complex tasks.
2026-05-19Multica Private Deployment Guide: Docker Compose, Daemon, Login, and Production Hardening (2026)
A practical guide to self-hosting multica-ai/multica on your own server or private network: Docker Compose quick install, manual setup, login codes, Caddy/Nginx reverse proxy, daemon pairing, Kubernetes, upgrades, and security boundaries.
2026-05-19Orkas Setup Guide: Local Desktop Commander + Workers Agent Team (2026)
Orkas is an MIT-licensed open-source desktop agent collaboration client with a unique Commander + Workers architecture. Covers installers, source setup, model configuration, built-in agents, external CLI backends, and conditional reflection.
2026-05-19Slock Setup Guide: Manage Your AI Agent Team via Chat Channels (2026)
Slock treats humans and AI agents as teammates in chat channels. This 10-minute guide walks you through: registering an account → starting the local daemon → creating your first agent → @-mentioning to assign work → understanding how task claim prevents file conflicts.
2026-05-19Conductor.build: Run a Team of Parallel AI Coding Agents on Your Mac
Conductor by Melty Labs is a Mac app that orchestrates parallel Claude Code and Codex agents, each working in an isolated git worktree. The app is free — you bring your own Claude or Codex subscription.
2026-05-16Choosing an agent collaboration platform: chat, task board, AI team workspace, or agent runtime
Agent collaboration platforms fall into four groups: chat, task board, AI team workspace, and agent runtime. The task-board group includes Multica and Orbi. Compare the interaction model before picking a product.
2026-05-092026 Agent Collaboration Platform Guide: Slock, Multica, LobeHub, Orkas Compared
From solo coding to agent teams — a deep comparison of four agent collaboration platforms: Slock's task claiming, Multica's skill sharing, LobeHub's agent marketplace, and Orkas's Commander dispatch. Find your ideal human+agent workflow.
2026-05-09AI Coding Tool Rate Limits Explained: How to Avoid Getting Blocked Mid-Sprint
Why does Claude Code, Cursor, or Copilot hit rate limits right when you're busiest? Learn how each tool's limits work and 5 practical strategies to keep working without interruption.
2026-04-05Essential VS Code Extensions for AI-Assisted Development
The best VS Code plugins to use alongside Cursor, Cline, and Claude Code — covering Git history, code quality, API testing, and readability. Build a complete AI-augmented workflow.
2026-04-01MCP Protocol Introduction: Connecting AI Coding Tools to the World
A comprehensive introduction to MCP (Model Context Protocol): what it is, why it matters, which AI coding tools support it, and how to configure MCP servers for Cursor, Claude Code, Cline, and more.
2026-03-28switch-ai: One Command to Switch AI API Configs — No More Manual Env Vars
Introducing switch-ai, a CLI tool for instantly switching between multiple AI API configurations (Anthropic, OpenAI, and more) using YAML profiles.
2026-03-02Tool Setup 5
Windsurf 2.0 Deep Dive: Full Evolution After the Cognition Acquisition
On April 15, 2026, Windsurf shipped its 2.0 major release. This guide covers the Cognition acquisition backstory, SWE-1.5 model, Codemaps visual codebase maps, Devin Cloud Agent integration, Agent Command Center, Spaces, and the complete pricing overhaul from monthly credits to daily/weekly quotas.
2026-05-25Cursor 3.0 Deep Dive: From Code Editor to Agent Command Center
On April 2, 2026, Cursor shipped its biggest redesign since launch — codename Glass. This guide covers the Agents Window, parallel agent execution, Design Mode, multi-repo support, and the May 13 cloud agent environments update.
2026-05-18Complete Guide to Using Cursor from China: Network, Payment, and Alternatives
Developers in China still need a proxy, a foreign card, and a plan for latency. This guide matches Cursor's 2026-10-16 pricing page: Free, Plus, and Premium, plus VPN-free alternatives.
2026-05-17Cursor Composer 2 Deep Dive: Kimi K2.5 Base + Reinforcement Learning, 37% Better
In March 2026 Cursor launched Composer 2 — a proprietary coding model trained on top of Moonshot AI's open-source Kimi K2.5. Here's the full story: the technology, the attribution controversy, and what it means for developers.
2026-04-20Trae CN Complete Getting Started Guide: Free AI IDE for China, Zero-Cost AI Coding
ByteDance's Trae CN is completely free, requires no VPN, and includes top Chinese AI models like Doubao, DeepSeek, and Kimi K2. This guide takes you from installation to SOLO mode hands-on.
2026-02-17Dev Practices 2
Grok Bot Architecture Explained: The Shared Computer, Task Loop, and Security Boundaries
Follow one concrete task through Grok Bot: a persistent cloud computer per user, bots that share it, and how plugins, shell / CLI, computer use, routines, and Auto Review divide the work. Includes diagrams and evidence limits.
2026-09-28Muse Architecture Explained: Agent Loops, Dedicated VMs, Memory, and Background Work
Follow a task through Meta Muse: Hatch advances the agent loop inside a dedicated VM, inference runs outside it, and context, memory, subagents, and permissions keep the work moving. Includes an architecture diagram and evidence limits.
2026-09-24Market Notes 4
AI Coding Agent Security Guide 2026: How to Protect Against Repository Access, Shell Execution, and Prompt Injection Risks
AI coding agents are getting more capable, but more power means more risk. This guide covers four critical dimensions — repository permissions, shell execution, prompt injection, and credential leaks — with practical security measures drawn from Claude Code, Codex, Copilot, Gemini CLI, and Cursor.
2026-08-13What Is a Meta-Harness? A 2026 Buyer's Guide to the Layer Above Your Coding Agents
Databricks Omnigent, Zed ACP, Vercel HarnessAgent, Cloudflare Flue, Conductor — a wave of meta-harnesses landed in one week of June 2026. A selection-focused survey: what it is, what it solves, how it relates to MCP/agent runtimes/collaboration platforms, and whether you should adopt one now.
2026-06-26Google AX, Agyn, and Dapr Agents Compared: Choosing an Agent Runtime
Google AX points to a new class of distributed agent runtime. This guide compares the two closest open-source projects, Agyn and Dapr Agents, across execution units, recovery, isolation, deployment, and fit.
2026-06-06AI Coding Agents in 2026: A Practical Roadmap from Autocomplete to Cloud Teammates
As of June 2026, AI coding tools are moving from autocomplete and AI IDEs into engineered agent workflows. This guide maps the shift across Codex, Claude Code, GitHub Copilot, Gemini CLI, MCP, security, and team adoption.
2026-06-01