Guides & fixes
Practical, no-fluff guides for people building with AI dev tools — fixing broken installs, working around lock-outs, and picking tools you can actually own.
Muse Spark 1.3 in OpenCode: Model IDs, Access and Privacy
Find Muse Spark 1.3 in the official OpenCode Zen catalog. Distinguish the standard model from Contributor Free and check data terms before using private code.
Read guide →OpenCode v2: Install Options and a V1 Migration Checklist
Check OpenCode v2 installation, shared configuration paths, plugin and server API changes, and a safe rollback before migrating from V1.
Read guide →DeepSeek Harness: What It Is and How to Try It Safely
Understand DeepSeek’s open-source agent harness, its local Web UI and developer-preview limitations before installing plugins or connecting a model.
Read guide →Codex Skills Explained (2026)
Codex skills extend the Codex CLI by adding reusable capabilities and workflows. What they are, how they work, and how to get more out of Codex.
Read guide →What Is Claude Cowork? (Anthropic's Collaboration Feature, Explained)
Claude Cowork is Anthropic's feature for working alongside Claude on shared, multi-step tasks rather than one-off prompts. What it does, who it's for, and how it compares.
Read guide →Windsurf vs Claude Code in 2026: An Honest Comparison
Windsurf is an AI-native IDE (now under Cognition) with agentic Cascade/Tab; Claude Code is a terminal agent on Anthropic models from $20/mo. Here's which fits your stack.
Read guide →Claude Code vs Codex in 2026: An Honest Comparison
On the independent Terminal-Bench 4.0 they tie (Codex 58.2% vs Claude Code 57.9%), and both start at $20/mo. The real split is ecosystem: Anthropic vs OpenAI.
Read guide →Cursor vs Claude Code vs Codex: An Honest 2026 Comparison
Cursor (AI editor), Claude Code and Codex (terminal agents), all from $20/mo. On Terminal-Bench 4.0, Codex and Claude Code tie (58.2% vs 57.9%). Here's how to pick.
Read guide →Best AI Coding Tools in 2026 (Tested & Ranked)
Claude Code, Codex, Cursor and Windsurf compared, all from $20/mo. Claude Code and Codex tie on Terminal-Bench 4.0 (57.9% vs 58.2%). Here's which fits your workflow.
Read guide →What Is t3 code? (2026 Guide)
t3 code is an open-source GUI from Theo Browne that manages AI coding agents (Codex, Claude, Cursor) in one place. What it does, and how it compares to a hosted router.
Read guide →Best Claude Code Plugins in 2026 (Tested & Ranked)
The best Claude Code plugins add testing, git, docs, and workflow skills. Our 2026 shortlist covers which marketplace categories are worth installing.
Read guide →What Are Claude Code Plugins? (Complete 2026 Guide)
Claude Code plugins extend the agent with new commands, skills, and integrations from a marketplace. What they are, how to install them, and which ones to start with.
Read guide →Google Antigravity vs Claude Code in 2026: An Honest Comparison
Antigravity is Google's AI coding agent; Claude Code is Anthropic's terminal agent. We compare them on workflow, models, and lock-in to help you pick the right one in 2026.
Read guide →What Is cc-switch? (2026 Guide)
cc-switch is a desktop app that manages AI coding CLIs (Claude Code, Codex, Grok) and switches providers in one click. How it works, and when a hosted router is simpler.
Read guide →Is Claude Unlimited? Plans, Usage Credits and API Limits
Claude plans have usage allowances. Extra credits and API billing can extend access, but rate limits, budgets and account policies still apply.
Read guide →Claude Code Subscription Access Disabled: Checks and Fixes
Check your organization, sign-in method and administrator policy when Claude Code subscription access is disabled. API billing is separate.
Read guide →What Is the Codex CLI? (2026 Guide)
The Codex CLI is OpenAI's terminal coding agent that edits files and runs commands from your shell. What it does, how to install it, and how to run it against other models.
Read guide →Is Codex Free? Free Plan Access, Limits and API Costs
Codex is included with ChatGPT Free and Go, with usage limits. Compare local access, Cloud eligibility and separate API billing before paying.
Read guide →Claude Code vs Cursor in 2026: An Honest Comparison
Cursor is an AI-native editor (free, Pro $20/mo, Ultra $200); Claude Code is a terminal agent on Anthropic models ($20/mo). One in your IDE, one in your shell.
Read guide →Best Claude Code Routers in 2026 (Tested & Compared)
The main difference between routers comes down to self-hosted vs managed. We compare OmniRoute, claude-code-router, AgentRouter, Cline and more for coding agents in 2026.
Read guide →What Is an Agent Router? (And Why Coding Agents Need One)
An agent router is one endpoint that lets Claude Code, Codex, or Cursor reach many model providers. How it works, when you need one, and open-source vs hosted options.
Read guide →Claude Tool-Use Limit: What It Means and What to Try
Claude stopped at a tool-use limit? Check the client and exact error, review completed work, then try a bounded continuation or a smaller task.
Read guide →Unlimited Claude API? Rate Limits, Budgets and Billing
Claude API usage is metered and rate-limited. Understand token billing, account budgets and retries before choosing an API route over a subscription.
Read guide →What Is OmniRoute? Free Local AI Gateway Explained (2026)
OmniRoute is a self-hosted AI gateway. Learn the provider-to-client setup, authentication checks and costs before routing a coding agent through it.
Read guide →How to use GPT-6.1 in Claude Code
Want Claude Code on OpenAI's GPT-6.1? GPT is OpenAI-compatible, so Claude Code needs a translating proxy. Here's the setup, and a cleaner, proxy-free way.
Read guide →How to use Claude Sonnet 5.5 in Claude Code
Claude Sonnet 5.5 is Anthropic's newest Sonnet, and Claude Code runs it natively, no proxy. How to select it, set the model, and a cleaner way to switch models.
Read guide →GPT-6 in Claude Code: Integration Checks and Limits
Check GPT-6 API and tool requirements, Claude Code gateway limitations and model costs before trying a third-party integration.
Read guide →How to use Claude Opus 5.5 in Claude Code
Claude Opus 5.5 is Anthropic's flagship, and Claude Code runs it natively, no proxy. How to select it, when it's worth the cost, and a cleaner way to switch models.
Read guide →How to use Grok 4.7 in Claude Code
Want Claude Code on xAI's Grok 4.7? Grok's API is OpenAI-compatible, so Claude Code needs a translating proxy. Here's the setup, and a cleaner option that skips it.
Read guide →GPT-6 Astra vs Claude Opus 5: which should you code with?
GPT-6 Astra vs Claude Opus 5 for coding agents: benchmarks, API pricing ($10/$50 vs $5/$25), context windows, and an honest take on which to use — plus how to keep both a click apart.
Read guide →GPT-6 Astra too expensive? Cheaper alternatives for coding agents
GPT-6 Astra is OpenAI's new flagship at $10/$50 per 1M tokens — twice Claude Opus 5. Here are cheaper, capable alternatives for coding agents (open-weight and hosted), and how to route your agent to them.
Read guide →DeepSeek V4.1 Flash: what's new, what it costs, and how to code with it
DeepSeek V4.1 Flash (Sept 2026) is a 552B MoE, natively multimodal, 1M-context open-weight model at rock-bottom prices — $0.15/$0.60 per 1M tokens. Here's what's new, what it costs, and how to run it in a coding agent.
Read guide →How to use GLM in Claude Code
Run Claude Code on Zhipu's GLM models, including the new GLM-5.3 and GLM-5.3-Flash (top open-weight coding model). GLM exposes an Anthropic-compatible endpoint, so Claude Code can point straight at it — here's the exact config, plus a cleaner one-click model switch.
Read guide →What is Qwen Code?
Qwen Code is Alibaba's open-source terminal coding agent, built around the Qwen3-Coder model. Here's what it is, how it compares to Claude Code and Gemini CLI, pricing, and a free GUI alternative if you don't live in the terminal.
Read guide →Codex Pricing: Free Access, Plan Limits and API Billing
Compare Codex access through ChatGPT Free and paid plans, optional credits and separately billed API keys. Check current prices before buying.
Read guide →Gemini 3.8 Flash in Claude Code: Compatibility Checks
Check Gemini API requirements, Claude Code gateway limitations and model costs. A base URL change alone does not verify a working integration.
Read guide →How to use DeepSeek-V4.1-Flash in Claude Code
DeepSeek-V4.1-Flash is fast and cheap, with an Anthropic-compatible endpoint, so Claude Code points straight at it, no proxy. The config, and the cleanest one-click way.
Read guide →How to use Kimi K3 in Claude Code
Kimi K3 from Moonshot has an Anthropic-compatible endpoint, so Claude Code points straight at it, no proxy. The exact config, and the cleanest one-click way to run it.
Read guide →How to use OpenRouter in Claude Code
OpenRouter puts hundreds of models behind one API key. Here's how to route Claude Code through OpenRouter, the gotchas, and a cleaner agent that switches models without a proxy.
Read guide →How to use Grok in Claude Code
Want Claude Code to run on xAI's Grok? Grok's API is OpenAI-compatible, so you route Claude Code through a translator or use an agent that speaks it natively. Here's the setup and the cleaner option.
Read guide →How to use Kimi in Claude Code
Run Claude Code on Moonshot's Kimi models. Kimi offers an Anthropic-compatible endpoint, so Claude Code can point straight at it. Here's the exact config — and a cleaner one-click model switch.
Read guide →How to use Qwen in Claude Code
Run Claude Code on Alibaba's Qwen3-Coder in 3 steps. DashScope is OpenAI- and Anthropic-compatible, so no proxy needed — copy-paste the exact config here, plus a free one-click alternative.
Read guide →How to use DeepSeek in Claude Code
Want Claude Code to run on DeepSeek instead of Claude? DeepSeek exposes an Anthropic-compatible endpoint, so you can point Claude Code straight at it. Here's the exact config — and the cleaner one-click option.
Read guide →Claude Code MCP not working: fix a server that won't connect
MCP server won't connect or show up in Claude Code? 90% of the time it's one of 4 things: invalid config JSON, wrong command path, missing env vars, or no restart. Here's the full fix checklist.
Read guide →Claude Code rate limit: what a 429 means and how to fix it
Hitting a rate limit or 429 too many requests in Claude Code or the Claude API. What a rate limit is, why coding agents trigger it, and how to back off, retry, and route around it.
Read guide →Claude Code "usage limit reached"? 6 fixes to keep working
You hit usage limit reached in Claude Code. Here is why subscription plans cap usage, what to do right now, and how to avoid it next time.
Read guide →Claude Code pricing: how the cost model actually works
Claude Code pricing in 2026: every subscription plan, pay-as-you-go API token rates, and the hidden costs that spike your bill on big codebases. Plus 5 ways to pay less and a free, open-source alternative.
Read guide →Cursor vs Claude Code
Cursor vs Claude Code compared for real work: editor vs terminal agent, pricing, model choice, and who each one suits. Plus a free, open-source option to try against both.
Read guide →OpenCode free models
How to use free models in OpenCode — where the free tiers come from, the limits and privacy trade-offs, and a free open-source agent that lets you plug in any free model with a proper GUI.
Read guide →Codex vs Claude Code
OpenAI Codex vs Anthropic's Claude Code — how the two terminal coding agents differ on models, cost, control, and lock-in, plus the free option when you don't want to be tied to either vendor.
Read guide →What is Ox Alpha?
Ox Alpha was a free, anonymous 1M-context coding model that shot to #1 on OpenRouter — now confirmed as Zhipu's GLM-5.3-Flash (320B-A18B, MIT). Here's what it is, how it performs, the privacy catch, and how to run it in a coding agent.
Read guide →Use Kimi or a local Ollama model in a coding agent
Want your coding agent to run on Kimi (Moonshot) or a local Ollama model instead of one fixed provider? Here's how each connects and what to check — plus an agent that makes switching a one-click setting.
Read guide →What is an agent router?
An agent router lets a coding agent switch between models/providers instead of being locked to one. Here's what it is, why people want it (cost, limits, outages), and how a built-in router beats bolting one on.
Read guide →How much does Claude Code cost — and how to spend less
Claude Code pricing explained: subscription vs API credits, why costs spike on big codebases, and concrete ways to spend less — including bringing your own cheaper model or running one locally for free.
Read guide →The best free, open-source Claude Code alternative
Want a free, open-source coding agent instead of Claude Code? One that runs locally, works with any model, and doesn't lock you to one provider. Here's what to compare, plus a straight answer on the best pick.
Read guide →Is Claude Code free?
Is Claude Code free? Short answer: no, it's tied to a paid plan or API usage. Here's what it really costs, whether there's a free tier, and a genuinely free, open-source alternative that runs locally.
Read guide →Claude Desktop connector says "connected" but files aren't showing?
Added a GitHub repo or folder to Claude Desktop, it says connected, but the file tree is empty and prompts can't read anything? Here's what's actually happening and how a local agent avoids the whole problem.
Read guide →OpenCode GUI: Desktop, Web and Alternative Apps
Compare OpenCode Desktop, Web and separate apps. Check installation, provider setup and model costs before choosing a graphical coding workflow.
Read guide →Can you use other models with Claude Code?
Short answer: yes, but it depends on how you set the endpoint. Here's how to point Claude Code at GLM, DeepSeek, or a local model — and the cleaner option when you just want an agent that swaps models natively.
Read guide →What is OpenClaw? The self-hosted personal AI assistant, explained
OpenClaw is an open-source, self-hosted personal AI assistant that runs on your own devices and meets you inside the chat apps you already use. Here's what it is, how it works, who it's for — and when you actually want a coding agent instead.
Read guide →Claude Desktop Won’t Open on Windows: Diagnose and Fix Launch Errors
Check the exact launch error, official installer, Windows repair options and event logs. Back up local data before reset; keep security updates enabled.
Read guide →Why desktop AI apps keep breaking — and what a local agent does differently
Self-corrupting installs, integrations that connect but do nothing, sessions that vanish — these aren't bugs in one app, they're what a hosted desktop AI app tends to become. Here's the pattern and how a local agent sidesteps it.
Read guide →How to use other models (GLM, DeepSeek, local) in a coding agent
Want to run a coding agent on GLM, DeepSeek, an OpenAI-compatible endpoint, or a local model instead of one fixed provider? Here's how model routing works and what to look for.
Read guide →Claude Code hit the rate limit or quota? Here's how to keep working
Claude Code says you've hit your usage limit or rate limit and stopped mid-task? Here's why it happens, how to keep working now, and how to stop it interrupting you again.
Read guide →Claude Code won't install or run on Windows? Here's how to fix it
Claude Code failing to install, 'command not found' after install, or crashing on start in Windows? Here are the usual causes — Node/npm, PATH, WSL, permissions — and how to get it running.
Read guide →Cursor won't open or stuck in a login loop? Here's how to actually fix it
Cursor won't launch, hangs on a white screen, or keeps kicking you back to the sign-in page? Here are the real causes and the fixes that hold — plus what to do when it keeps happening.
Read guide →Codex Won’t Open on Windows: Diagnose and Fix Launch Errors
Check the exact launch error, official installer, Windows repair options and event logs. Back up local data before reset; keep security updates enabled.
Read guide →