# guides-ai.com

> 152 short, numbered, copy-paste guides for AI tools — Claude Code, ChatGPT, Cursor, MCP servers,
> local models (Ollama, LM Studio), prompting, and AI image tools. Each guide ends in a command or exact
> steps you can copy. Guides are numbered NN.MM: NN = topic, MM = position inside the topic.

Search index (JSON): https://guides-ai.com/search-index.json
Full text of every guide (one file): https://guides-ai.com/llms-full.txt
Any guide as Markdown: https://guides-ai.com/guides/<slug>.md  (e.g. https://guides-ai.com/guides/run-local-llm-ollama-windows.md)
MCP server — connect any AI agent: https://guides-ai.com/mcp  (tools: search_guides, get_guide, list_guides, list_topics)
New & updated feed (RSS): https://guides-ai.com/rss.xml
Log of additions: https://guides-ai.com/log/

## Topics

- 01 [Install & set up](https://guides-ai.com/topics/setup/) — Get Claude Code installed, updated, and unstuck — Windows, macOS, Linux, VS Code. (7 guides)
- 02 [First steps](https://guides-ai.com/topics/first-steps/) — Your first session, plan mode, the shortcuts that matter, and Gemini CLI. (4 guides)
- 03 [Claude Code commands](https://guides-ai.com/topics/commands/) — Hooks, subagents, slash commands, CLAUDE.md, permissions, context, and costs. (14 guides)
- 04 [Workflows](https://guides-ai.com/topics/workflow/) — Reviewing PRs, working across a whole codebase, giving an agent the right context. (7 guides)
- 05 [MCP](https://guides-ai.com/topics/mcp/) — Install servers, connect GitHub, and the best servers to start with. (9 guides)
- 06 [Claude](https://guides-ai.com/topics/claude/) — Projects and Artifacts in the Claude app. (2 guides)
- 07 [ChatGPT](https://guides-ai.com/topics/chatgpt/) — Custom GPTs, Projects, custom instructions, voice mode, and long PDFs. (6 guides)
- 08 [Cursor](https://guides-ai.com/topics/cursor/) — AI coding inside the Cursor editor. (4 guides)
- 09 [Local LLMs](https://guides-ai.com/topics/local-llm/) — Ollama and LM Studio: run a model locally, pick a quantization, call the API. (19 guides)
- 10 [Prompting](https://guides-ai.com/topics/prompting/) — Better prompts, system prompts, chaining, RAG, and fewer hallucinations. (9 guides)
- 11 [Choosing tools](https://guides-ai.com/topics/choosing/) — Claude vs ChatGPT, Claude Code vs Cursor, which model, which API. (7 guides)
- 12 [AI media](https://guides-ai.com/topics/media/) — Generate and edit images, video and audio locally — FLUX, Midjourney, inpainting, background removal, upscaling, text to speech, transcription, OCR. (14 guides)
- 13 [Everyday tasks](https://guides-ai.com/topics/tasks/) — Practical jobs: spreadsheets, transcripts, summaries, data analysis with AI. (4 guides)
- 14 [Other coding agents](https://guides-ai.com/topics/agents/) — Codex CLI, Aider, Copilot CLI, Windsurf — the first session with each. (11 guides)
- 15 [Building with the API](https://guides-ai.com/topics/api/) — Claude and OpenAI APIs: first request, tool use, structured output, caching, batch, streaming. (23 guides)
- 16 [Hugging Face](https://guides-ai.com/topics/huggingface/) — The Hub in practice: download and cache models, gated repos, GGUF with Ollama, transformers, Inference Providers, datasets, uploads, Spaces. (12 guides)

## Guides

### 01 Install & set up

- 01.01 [Install Claude Code on Windows: Download, Setup, Run](https://guides-ai.com/guides/install-claude-code-windows/): irm https://claude.ai/install.ps1 | iex installs Claude Code on Windows 10 1809+. WinGet and npm alternatives, Git Bash setup, and the first login.
- 01.02 [How to Install Claude Code on macOS & Linux](https://guides-ai.com/guides/install-claude-code-macos-linux/): Install Claude Code with the curl script or Homebrew in under a minute. The npm route now needs Node.js 22+, and a paid Claude plan is required.
- 01.03 [-bash: claude: command not found? 4 Fixes That Work](https://guides-ai.com/guides/fix-claude-command-not-found/): claude: command not found means the install directory is not on your PATH. Four fixes: new terminal, PATH entry, duplicate install, npm repair.
- 01.04 [How to Use Claude Code in VS Code](https://guides-ai.com/guides/claude-code-vscode-extension/): Install the Claude Code extension from the VS Code Marketplace for a native panel with inline diffs and rewind — it no longer auto-installs itself.
- 01.05 [How to Update Claude Code (and Unstick It)](https://guides-ai.com/guides/update-claude-code/): claude update applies the newest Claude Code now, claude doctor says why one stalled, and package-manager installs need upgrading by hand. Copy-paste.
- 01.06 [Run Claude Code on Windows with WSL](https://guides-ai.com/guides/claude-code-wsl-windows/): Enable WSL with wsl --install, run the Linux installer for Claude Code inside Ubuntu, fix the PATH, and keep projects on the Linux filesystem.
- 01.07 [Use Claude Code in JetBrains IDEs](https://guides-ai.com/guides/claude-code-jetbrains/): Install the Claude Code [Beta] plugin from the JetBrains Marketplace, run claude in the IDE terminal, and connect external sessions with /ide.

### 02 First steps

- 02.01 [Your First Claude Code Session in 5 Minutes](https://guides-ai.com/guides/first-claude-code-session/): Run claude inside your project, give it one clear task, and check the permission mode first — auto mode may edit files without asking on paid plans.
- 02.02 [How to Use Plan Mode in Claude Code (and When)](https://guides-ai.com/guides/claude-code-plan-mode/): Plan mode makes Claude Code research and propose before editing. Enter it with Shift+Tab or /plan, review the plan, then approve — or keep planning.
- 02.03 [Claude Code Keyboard Shortcuts & Terminal Basics](https://guides-ai.com/guides/claude-code-terminal-shortcuts/): The keys that make Claude Code fast: @ for files, ! for shell, Esc to interrupt, Esc Esc to rewind, Shift+Tab for modes, Ctrl+B to background.
- 02.04 [Getting Started with Gemini CLI](https://guides-ai.com/guides/gemini-cli-first-steps/): Install Gemini CLI with npm, npx or Homebrew, sign in with Google for 60 requests a minute free, and run your first prompt inside a project.

### 03 Claude Code commands

- 03.01 [How to Set Up a CLAUDE.md That Actually Helps](https://guides-ai.com/guides/create-claude-md-project-memory/): Run /init, keep CLAUDE.md under 200 lines, and confirm it loaded with /context. Plus .claude/rules for path-scoped instructions and @path imports.
- 03.02 [Create a Custom Slash Command in Claude Code](https://guides-ai.com/guides/create-claude-code-slash-command/): Drop a Markdown file in .claude/commands/ and the filename becomes the command. $0 is the first positional argument, and skills are now preferred.
- 03.03 [How to Configure Hooks in Claude Code](https://guides-ai.com/guides/claude-code-hooks/): Hooks run a shell command on Claude Code events. Note $CLAUDE_FILE_PATHS no longer exists — read the file path from the JSON on stdin instead.
- 03.04 [How to Use Subagents in Claude Code](https://guides-ai.com/guides/claude-code-subagents/): A subagent is a separate Claude with its own context window. Write one as a Markdown file in .claude/agents/ — note /agents is no longer a wizard.
- 03.05 [Manage Context in Claude Code: /clear, /compact](https://guides-ai.com/guides/claude-code-manage-context/): Use /clear between tasks, /compact within one, /context to see what fills the window, and claude -c to resume. Keeps long sessions fast and cheap.
- 03.06 [How to Track and Cut Claude Code Token Usage](https://guides-ai.com/guides/claude-code-track-costs/): Run /usage to see session tokens, cost and plan limits — /cost is now just an alias. Then cut waste with /clear, subagents and the right model.
- 03.07 [Control What Claude Code Can Do: Permissions](https://guides-ai.com/guides/claude-code-permissions/): Claude Code now starts in auto mode on Pro, Max and Team plans. Shift+Tab to switch, and write allow/deny rules that actually match the command.
- 03.08 [Claude Code settings.json: Which File, Which Key](https://guides-ai.com/guides/claude-code-settings-json/): The four settings files Claude Code reads, which one wins when they disagree, and the handful of keys worth setting on day one.
- 03.09 [Claude Code Skills and Plugins](https://guides-ai.com/guides/claude-code-skills-and-plugins/): Skills are folders with a SKILL.md. Plugins bundle skills, agents and hooks for sharing. Install one from a marketplace, then write your own.
- 03.10 [Claude Code usage limits and how to track them](https://guides-ai.com/guides/claude-code-usage-limits/): What session, weekly, and per-model limits mean on Pro and Max, how /usage shows your plan usage, and what to do when you hit a limit.
- 03.11 [Pick the right Claude model in Claude Code](https://guides-ai.com/guides/claude-code-choose-a-model/): Switch between Opus, Sonnet, and Haiku with /model, set a default, and match the model to the job without guessing at prices or benchmarks.
- 03.12 [Claude Code memory: the CLAUDE.md hierarchy and /memory](https://guides-ai.com/guides/claude-code-memory-hierarchy/): Claude Code concatenates CLAUDE.md from managed policy, your home dir, the project, then CLAUDE.local.md. Inspect it with /memory and @path imports.
- 03.13 [Change how Claude Code responds with output styles](https://guides-ai.com/guides/claude-code-output-styles/): List, switch, and build output styles in Claude Code with /output-style, and see how a style differs from CLAUDE.md and the system prompt.
- 03.14 [Fix Claude Code usage/rate limit reached](https://guides-ai.com/guides/fix-claude-code-rate-limit/): Hit a usage or rate limit in Claude Code? What the session, weekly, and per-model limits mean, and the fastest ways to get working again.

### 04 Workflows

- 04.01 [Let Claude Code Work on a Whole Repo Safely](https://guides-ai.com/guides/claude-code-whole-codebase-safely/): Branch, plan, deny the destructive commands, review in small chunks. Auto mode now starts on by default, so don't assume Claude will stop and ask.
- 04.02 [Give an AI Coding Agent the Right Context](https://guides-ai.com/guides/ai-coding-agent-context/): CLAUDE.md, AGENTS.md and .cursor/rules: four habits that fix most bad AI code — a rules file, @file references, one example, a clean context.
- 04.03 [How to Review a Pull Request with Claude Code](https://guides-ai.com/guides/claude-code-review-a-pr/): Run /code-review (alias /review) on a diff, or check out a PR with gh. Ask for correctness over style, then have Claude apply only fixes you accept.
- 04.04 [How to Set Up the Claude Code GitHub Action](https://guides-ai.com/guides/claude-code-github-actions/): Install the Claude GitHub App, add one secret, and drop in a workflow so @claude answers in your issues and PRs. Copy-paste.
- 04.05 [How to Run Claude Code Headless in a Script or CI](https://guides-ai.com/guides/claude-code-headless-ci/): Run Claude Code non-interactively with claude -p, get JSON back, pre-approve tools, and wire it into a CI job. Copy-paste.
- 04.06 [Run Parallel Claude Code Sessions with Git Worktrees](https://guides-ai.com/guides/claude-code-parallel-sessions-worktrees/): Start a second Claude Code session in its own git worktree with `claude --worktree`, keep .env files in it, and clean up when you're done. Copy-paste.
- 04.07 [Run Claude Code in a Docker devcontainer](https://guides-ai.com/guides/claude-code-docker-devcontainer/): Add Anthropic's official dev container feature to .devcontainer/devcontainer.json, rebuild, and run claude inside an isolated, reproducible container.

### 05 MCP

- 05.01 [How to Install an MCP Server in Claude Code](https://guides-ai.com/guides/install-mcp-server-claude-code/): Add an MCP server with claude mcp add: -- for local commands, --transport http for hosted ones, --scope to choose who gets it, then /mcp to confirm.
- 05.02 [Connect Claude Code to GitHub with MCP](https://guides-ai.com/guides/connect-claude-code-to-github/): Add GitHub's hosted MCP server at api.githubcopilot.com with a bearer token — the old @modelcontextprotocol/server-github package has been archived.
- 05.03 [The Best MCP Servers to Start With](https://guides-ai.com/guides/best-mcp-servers-to-start/): Seven MCP reference servers are still maintained — filesystem, fetch, git, memory, time. GitHub, Postgres, Slack and Puppeteer were archived.
- 05.04 [MCP in Claude Code vs Claude Desktop](https://guides-ai.com/guides/mcp-claude-code-vs-desktop/): claude mcp add registers a server from the terminal; Claude Desktop edits a JSON config from Settings. Same protocol, two setup flows, a scope trick.
- 05.05 [How to Build an MCP Server in Python](https://guides-ai.com/guides/build-mcp-server-python/): Write a one-tool MCP server named word_count with the official Python SDK, run it over stdio, and call it from Claude Code with claude mcp add.
- 05.06 [How to Build an MCP Server in TypeScript](https://guides-ai.com/guides/build-mcp-server-typescript/): Write a one-tool MCP server with the official TypeScript SDK, serve it over stdio, and call it from Claude Code. Copy-paste.
- 05.07 [MCP Server Not Working in Claude Code?](https://guides-ai.com/guides/claude-code-mcp-troubleshooting/): claude mcp list says Failed to connect or Pending approval? Five checks find the cause: scope, approval, the -- separator, env vars and timeouts.
- 05.08 [Connect a Remote MCP Server to Claude Code over HTTP](https://guides-ai.com/guides/claude-code-remote-mcp-servers/): Add a hosted MCP server with claude mcp add --transport http, authenticate with OAuth or a header, and pick the right scope for your team.
- 05.09 [Connect the guides-ai MCP Server to Your Agent](https://guides-ai.com/guides/connect-guides-ai-mcp-server/): Add the guides-ai MCP server to Claude Code, Cursor, or any agent with one URL. No install, no token. Then search the whole guide library from chat.

### 06 Claude

- 06.01 [How to Use Claude Projects](https://guides-ai.com/guides/claude-projects-getting-started/): Claude Projects keep files and instructions across chats. Free accounts get up to 5 projects; paid plans add RAG that expands knowledge up to 10x.
- 06.02 [How to Use Claude Artifacts](https://guides-ai.com/guides/claude-artifacts/): Artifacts open code, docs and diagrams in a side panel. Turn on Code execution and file creation in Settings, iterate in place, then publish a link.

### 07 ChatGPT

- 07.01 [ChatGPT PDF Summary: Summarize Long Files in 3 Steps](https://guides-ai.com/guides/chatgpt-summarize-pdf/): Attach the PDF in ChatGPT, ask for a shape instead of a summary, then drill in. The 512 MB and 2M-token caps, and what to do with a scanned file.
- 07.02 [How to Set ChatGPT Custom Instructions](https://guides-ai.com/guides/chatgpt-custom-instructions/): Custom instructions moved to Settings then Personalization and are now one field, not two boxes. They apply immediately to all chats, on every plan.
- 07.03 [How to Build a Custom GPT (No Code)](https://guides-ai.com/guides/build-a-custom-gpt/): Custom GPT creation now needs a Business, Enterprise or Edu workspace. Use the Configure tab for instructions, up to 20 knowledge files, and actions.
- 07.04 [How to Use ChatGPT Projects](https://guides-ai.com/guides/chatgpt-projects/): ChatGPT Projects group chats around shared files and instructions. Free plans get 5 files per project, Plus 25, Pro and Business 40; sharing is free.
- 07.05 [How to Use ChatGPT Voice Mode](https://guides-ai.com/guides/chatgpt-voice-mode/): ChatGPT voice now runs inside the chat, not a separate blue-orb screen. Tap the Voice icon on web, hold the mic on mobile, and when to skip it.
- 07.06 [How to Get Working Excel Formulas from ChatGPT](https://guides-ai.com/guides/excel-formulas-with-chatgpt/): Describe the sheet so ChatGPT can answer, ask for the formula plus a breakdown, verify it on real rows, and fix the four things that go wrong.

### 08 Cursor

- 08.01 [How to Use Cursor: A 10-Minute First-Edit Tutorial](https://guides-ai.com/guides/cursor-first-steps/): Cursor in ten minutes: Tab completions, Cmd/Ctrl K inline edits, the Agent sidepane on Cmd/Ctrl I, and how to review or undo what Agent wrote.
- 08.02 [Add an MCP Server to Cursor (.cursor/mcp.json)](https://guides-ai.com/guides/cursor-mcp-servers/): Configure a local or remote MCP server in Cursor, global or per project, keep the API key out of git, and verify it with a real tool call.
- 08.03 [How to Write a Cursor Project Rule (.cursor/rules)](https://guides-ai.com/guides/cursor-rules-file/): Cursor project rules are .mdc files with frontmatter. Learn the four rule types, which frontmatter produces each, and copy a working example.
- 08.04 [Run Cursor Background (Cloud) Agents](https://guides-ai.com/guides/cursor-background-agents/): Cursor renamed Background Agents to Cloud Agents. Run async agents in a cloud VM on your repo, review their PRs, and learn the privacy limits.

### 09 Local LLMs

- 09.01 [How to Run Ollama on Windows (Install, GPU, Models)](https://guides-ai.com/guides/run-local-llm-ollama-windows/): Install Ollama on Windows 10 22H2+, run your first model with ollama run, check it landed on the GPU, and move the model store off your C: drive.
- 09.02 [Ollama Modelfile: a Custom Model With a System Prompt](https://guides-ai.com/guides/ollama-custom-model-modelfile/): An Ollama Modelfile bakes FROM, SYSTEM and PARAMETER into a named local model with ollama create, so behaviour travels with the model, not scripts.
- 09.03 [Call Your Local Ollama Model from curl & Python](https://guides-ai.com/guides/call-ollama-api/): Ollama serves a local API on port 11434. Call /api/chat with curl or Python, or point any OpenAI client at http://localhost:11434/v1/ instead.
- 09.04 [Choose a GGUF Quantization: Q4 vs Q5 vs Q8](https://guides-ai.com/guides/choose-llm-quantization/): Q4_K_M is the default pick for local LLMs. Real bits-per-weight for each GGUF level, why Q8_0 is a legacy format, and how to match one to your RAM.
- 09.05 [Run a Local LLM with LM Studio (No Terminal)](https://guides-ai.com/guides/run-llm-lm-studio/): LM Studio downloads and runs GGUF models in a desktop app, then serves them at http://localhost:1234/v1 for any OpenAI-compatible client. No terminal.
- 09.06 [llama.cpp Quickstart: Run a GGUF Model](https://guides-ai.com/guides/llama-cpp-quickstart/): Install llama.cpp, pull a GGUF from Hugging Face with llama cli -hf, chat in the terminal, then serve an OpenAI-compatible API on localhost:9931.
- 09.07 [How Much RAM or VRAM a Local LLM Needs (1B to 70B)](https://guides-ai.com/guides/local-llm-ram-vram-requirements/): Real GGUF file sizes at Q4/Q8 for 1B to 70B models, the rule of thumb behind them, and what fits in 8, 16, 24 or 64 GB. Pick a model that runs.
- 09.08 [How to Run Ollama in Docker with an NVIDIA GPU](https://guides-ai.com/guides/ollama-in-docker-with-gpu/): NVIDIA Container Toolkit setup, the official docker run command, pulling a model, keeping models in a volume, and AMD or CPU fallbacks.
- 09.09 [How to Run Open WebUI with Ollama (Docker)](https://guides-ai.com/guides/open-webui-with-ollama/): Give your local Ollama models a ChatGPT-style web interface with one Docker command — first login, model picker, and fixes.
- 09.10 [How to Run a Local LLM with Ollama on macOS/Linux](https://guides-ai.com/guides/run-local-llm-ollama-macos-linux/): Install Ollama with one command, pull gemma3:4b, chat offline in the terminal, and keep the daemon running as a background service on port 11434.
- 09.11 [How to Serve a Model with vLLM](https://guides-ai.com/guides/vllm-serve-openai-compatible/): Install vLLM, start a server with vllm serve on localhost:8000, and call it with the OpenAI Python SDK. Needs a Linux box with an NVIDIA GPU.
- 09.12 [Run LLMs Fast on Apple Silicon with MLX](https://guides-ai.com/guides/local-llm-apple-silicon-mlx/): MLX is Apple's framework for M-series Macs. Install mlx-lm, generate from a Hugging Face model, and serve an OpenAI-style API with mlx_lm.server.
- 09.13 [Fix Ollama Not Using Your GPU](https://guides-ai.com/guides/ollama-gpu-not-detected/): Ollama ran on CPU instead of your GPU. Confirm it with ollama ps, then fix the driver, Docker and WSL passthrough, or a model too big for VRAM.
- 09.14 [Call any local model with the OpenAI-compatible API](https://guides-ai.com/guides/call-a-local-model-openai-compatible/): Ollama, LM Studio, llama.cpp and vLLM all expose an OpenAI-compatible /v1 API. List models, run a chat completion, and stream from one SDK.
- 09.15 [Fix CUDA out of memory when running a model locally](https://guides-ai.com/guides/fix-cuda-out-of-memory/): torch.OutOfMemoryError: CUDA out of memory means the GPU is full. Lower the batch size, use float16 or 4-bit, offload to CPU, or clear the cache.
- 09.16 [Fix Ollama model not found or pull failed](https://guides-ai.com/guides/fix-ollama-model-not-found/): ollama run says model not found? Pull the exact model:tag first, fix a wrong or case-sensitive tag, or run a GGUF straight from Hugging Face.
- 09.17 [Fix Ollama out of memory or model too large](https://guides-ai.com/guides/fix-ollama-out-of-memory/): Ollama ran out of memory or the model is too large. Check the split with ollama ps, then cut the model, the quant, or the context window.
- 09.18 [Fix pip install torch (CPU-only or CUDA mismatch)](https://guides-ai.com/guides/fix-pip-install-torch-cuda/): Plain pip install torch gives a CPU-only build on Windows. Install from the PyTorch index URL that matches your CUDA, then verify the GPU.
- 09.19 [Use a Local Model in Cursor or Claude Code](https://guides-ai.com/guides/local-model-in-cursor-or-claude-code/): Point Cursor's chat at a local OpenAI-compatible model, and see why Claude Code cannot use a non-Claude or local model as its backend.

### 10 Prompting

- 10.01 [How to Write Better AI Prompts (A Framework)](https://guides-ai.com/guides/write-better-ai-prompts/): A prompt framework that still holds: role, task, context, format — plus the long-input ordering rule Anthropic measures at up to 30% better answers.
- 10.02 [How to Write a System Prompt That Steers a Model](https://guides-ai.com/guides/write-a-system-prompt/): System prompt structure: role, rules with reasons, exact output format, tagged examples. Set it in the API system parameter, then test by breaking it.
- 10.03 [How to Reduce AI Hallucinations With Prompts](https://guides-ai.com/guides/reduce-ai-hallucinations/): Ground answers in pasted source text, allow "not stated", demand a quote per claim, and lower temperature: the habits that cut invented facts.
- 10.04 [How to Chain Prompts to Solve a Big Task](https://guides-ai.com/guides/prompt-chaining/): Prompt chaining splits a task into separate API calls you can inspect (draft, review, refine), and when a modern model handles it in one call.
- 10.05 [RAG Explained Simply (and When You Need It)](https://guides-ai.com/guides/rag-explained-simply/): RAG retrieves relevant chunks, then generates. With 1M-token context windows and hosted file_search tools, here is when it beats pasting the file.
- 10.06 [Prompt Evals with promptfoo: One YAML, One Matrix](https://guides-ai.com/guides/evals-with-promptfoo/): Install promptfoo, write a config with two prompts, three test cases and real assertions, run it, and read the pass/fail grid.
- 10.07 [LLM-as-Judge Evals: Rubric and Verdict](https://guides-ai.com/guides/llm-as-judge-evals/): Grade model outputs with a judge model: a copy-paste rubric prompt, a structured verdict, a Python scoring loop, and the biases that fake good scores.
- 10.08 [Prompt Injection Defense for Tool-Using Agents](https://guides-ai.com/guides/prompt-injection-defense/): Six defenses that actually reduce blast radius — privilege separation, allowlists, fenced tool results, approval gates, output filtering, and tests.
- 10.09 [Generate Embeddings Locally with EmbeddingGemma](https://guides-ai.com/guides/local-embeddings-embeddinggemma/): Install sentence-transformers, load EmbeddingGemma 2, encode queries and documents with its prefixes, score cosine similarity, and feed a vector DB.

### 11 Choosing tools

- 11.01 [Which AI Model Should You Use?](https://guides-ai.com/guides/pick-an-ai-model/): Pick an AI model by job: coding, images, very long inputs, or private local inference. Current tiers, with the context-window myth corrected.
- 11.02 [Claude vs ChatGPT for Coding: Which to Use](https://guides-ai.com/guides/claude-vs-chatgpt-for-coding/): A practical split for real coding work, plus the deciding fact most comparisons skip: Claude Code needs a paid plan, ChatGPT answers on free.
- 11.03 [Claude Code vs Cursor: Which One to Use](https://guides-ai.com/guides/claude-code-vs-cursor/): Claude Code is a terminal agent; Cursor is an agent-first editor. Where each one wins, and why the Claude Code extension installs into Cursor too.
- 11.04 [OpenAI API vs Claude API: Which to Build On](https://guides-ai.com/guides/openai-vs-claude-api/): OpenAI and Claude API compared on price per million tokens, 1M-token context, tool use and long-context surcharges, plus how to decide with an eval.
- 11.05 [Fine-Tuning vs RAG vs Prompting: How to Choose](https://guides-ai.com/guides/fine-tuning-vs-rag-vs-prompting/): A decision table plus rules of thumb — which one fixes your problem, and what each costs in money, latency and maintenance.
- 11.06 [Ollama vs LM Studio: Which Local LLM Runner](https://guides-ai.com/guides/ollama-vs-lm-studio/): Ollama is a scriptable CLI and daemon; LM Studio is a desktop app with a model browser. Both run GGUF locally, so pick by workflow, not quality.
- 11.07 [Which Local LLM Should You Run?](https://guides-ai.com/guides/which-local-model-should-i-run/): Choose a local LLM by what fits your RAM and VRAM first, then by task, with starter picks for 8GB, 16GB, and 24GB machines.

### 12 AI media

- 12.01 [Remove an Image Background With AI (Free)](https://guides-ai.com/guides/remove-background-with-ai/): rembg cuts a subject to a transparent PNG from the command line — one file or a whole folder, offline, full resolution, no watermark and no upload.
- 12.02 [Midjourney Prompts That Actually Work](https://guides-ai.com/guides/midjourney-prompts-that-work/): Midjourney prompt structure: subject, style, details, then parameters last. --raw has replaced --style raw, and --ar still sets the aspect ratio.
- 12.03 [How to Inpaint: Edit One Part of an Image](https://guides-ai.com/guides/ai-image-inpainting/): Inpainting masks a region and regenerates only that: Midjourney Vary (Region), Firefly Generative Fill, ChatGPT, or Stable Diffusion locally.
- 12.04 [How to Install ComfyUI and Run Your First Workflow](https://guides-ai.com/guides/comfyui-first-workflow/): Install ComfyUI the reliable way, put a checkpoint in the right folder, load the default text-to-image graph, and generate an image.
- 12.05 [How to Generate Images Locally with FLUX (Python)](https://guides-ai.com/guides/generate-images-with-flux-locally/): Run FLUX.1-schnell locally on your GPU with a short Python diffusers script — accept the Hugging Face license and offload for low VRAM.
- 12.06 [How to Generate a Short AI Video Locally (Wan 2.2)](https://guides-ai.com/guides/generate-short-ai-video-locally/): The honest local text-to-video path: Wan 2.2 TI2V-5B in ComfyUI — three model files, real VRAM numbers, and a five-second clip.
- 12.07 [How to Generate Speech Locally with Piper TTS](https://guides-ai.com/guides/text-to-speech-locally/): Turn a text file into a WAV on your own machine — install Piper, download a voice, and choose between quality tiers and languages.
- 12.08 [How to Translate an SRT Subtitle File with AI](https://guides-ai.com/guides/translate-subtitles-with-ai/): Translate subtitles without breaking timestamps — a copy-paste Python script, the prompt that keeps line counts, and the checks that catch drift.
- 12.09 [How to Upscale Images Locally with Real-ESRGAN](https://guides-ai.com/guides/upscale-images-locally/): Enlarge photos and anime art 4x on your own machine with the standalone Real-ESRGAN binary — one command, no Python, and batch folders.
- 12.10 [How to Transcribe Audio Locally with Whisper](https://guides-ai.com/guides/whisper-transcribe-audio-locally/): Install OpenAI Whisper and ffmpeg, run whisper file.mp3 --model turbo, and get SRT and TXT subtitle files with timestamps out on your own machine.
- 12.11 [Clone a Voice Locally with Chatterbox TTS](https://guides-ai.com/guides/clone-a-voice-locally/): Use ResembleAI Chatterbox (MIT) to clone a voice from a short reference clip in a few lines of Python, with the consent caveat up front.
- 12.12 [Describe and Query Images with a Local VLM](https://guides-ai.com/guides/describe-images-with-a-vision-model/): Load Qwen3-VL-8B-Instruct with transformers, pass an image plus a question, and get captions, chart reads and answers on your own GPU.
- 12.13 [Generate Video Locally with LTX (diffusers)](https://guides-ai.com/guides/generate-video-with-ltx-locally/): Run Lightricks LTX-Video on your own GPU with a short diffusers script: LTXPipeline, text-to-video, frames and steps, plus the ComfyUI route.
- 12.14 [Extract Text from PDFs and Images with Local OCR](https://guides-ai.com/guides/run-ocr-locally/): Install indra-ocr, run a CPU-only OCR model on an image, pull Markdown from a multi-page PDF, batch files, and find where the weights cache.

### 13 Everyday tasks

- 13.01 [Clean Up Messy Spreadsheet Data with ChatGPT](https://guides-ai.com/guides/chatgpt-clean-spreadsheet-data/): Upload the CSV to ChatGPT data analysis, describe the mess once, and get back a cleaned file plus a repeatable formula or script. Limits and traps.
- 13.02 [How to Analyze a CSV with ChatGPT Data Analysis](https://guides-ai.com/guides/chatgpt-analyze-csv-data/): Upload a spreadsheet, get it cleaned, charted, and summarized — plus the limits worth knowing before you trust the numbers.
- 13.03 [How to Summarize a PDF Locally with Ollama](https://guides-ai.com/guides/summarize-pdf-locally-with-ollama/): Extract PDF text with pypdf, chunk it, summarize each chunk through the local Ollama API, and merge — about 30 lines, nothing uploaded.
- 13.04 [How to Summarize a YouTube Video with AI](https://guides-ai.com/guides/summarize-youtube-video-with-ai/): Pull the transcript with yt-dlp --convert-subs srt, then hand the .srt file to Claude or ChatGPT with a prompt that gives you timestamps, not mush.

### 14 Other coding agents

- 14.01 [How to Get Started with OpenAI Codex CLI](https://guides-ai.com/guides/codex-cli-first-steps/): Install Codex CLI with npm or Homebrew, sign in with ChatGPT, run your first task in a repo, and use /permissions to set what it can do. Copy-paste.
- 14.02 [How to Start Using Aider](https://guides-ai.com/guides/aider-first-steps/): Install Aider, point it at Claude or GPT with an API key, make your first edit, and control its git auto-commits. Copy-paste.
- 14.03 [How to Use GitHub Copilot CLI in Your Terminal](https://guides-ai.com/guides/github-copilot-cli-first-steps/): Install GitHub Copilot CLI, log in, ask it for a command or a code change, and control what it's allowed to run. Copy-paste.
- 14.04 [How to Start with Windsurf (Now Devin Desktop)](https://guides-ai.com/guides/windsurf-first-steps/): Windsurf is now Devin Desktop. Install it, open a project, run a first agent task, and find where your rules live. Copy-paste.
- 14.05 [Gemini CLI: Add MCP Servers and Install Extensions](https://guides-ai.com/guides/gemini-cli-mcp-and-extensions/): Wire an MCP server into Gemini CLI's settings.json or add it with one command, check it with /mcp, and install a packaged extension.
- 14.06 [AGENTS.md: one instructions file for every coding agent](https://guides-ai.com/guides/agents-md-file/): AGENTS.md is the cross-agent conventions file. Where to put it, what to write, which tools read it, and how Claude Code loads it next to CLAUDE.md.
- 14.07 [OpenCode: First Session with the Coding Agent](https://guides-ai.com/guides/opencode-first-steps/): Install the OpenCode CLI, connect a cloud or local Ollama model, open a project, give the agent a task, and compare it to Claude Code and Cursor.
- 14.08 [Add MCP servers to Codex CLI](https://guides-ai.com/guides/codex-cli-mcp/): Enable MCP servers in OpenAI Codex CLI: config.toml location, the exact mcp_servers TOML shape, a worked example, and how to confirm it loaded.
- 14.09 [AI coding agents compared: Claude Code, Codex, Cursor](https://guides-ai.com/guides/coding-agents-compared/): Compare Claude Code, Codex CLI, Copilot CLI, Cursor, OpenCode and Windsurf: terminal vs IDE, provider lock-in, MCP, open source and cost.
- 14.10 [Getting started with Devin (Cognition)](https://guides-ai.com/guides/devin-desktop-first-steps/): Access Devin by Cognition: the cloud web app, Devin Desktop IDE (formerly Windsurf), CLI and Slack, plus how it differs from terminal agents.
- 14.11 [Authenticate GitHub Copilot CLI (login and GITHUB_TOKEN)](https://guides-ai.com/guides/github-copilot-cli-authentication/): Log in to GitHub Copilot CLI with /login, check your Copilot subscription, and set GITHUB_TOKEN for CI with the Copilot Requests permission.

### 15 Building with the API

- 15.01 [Your First Claude API Request: curl, Python, Node](https://guides-ai.com/guides/claude-api-first-request/): Get a key, set ANTHROPIC_API_KEY, and make a working Messages API call in curl, Python, and Node — then read the response properly.
- 15.02 [Claude API Tool Use](https://guides-ai.com/guides/claude-api-tool-use/): Define a tool with a JSON schema, catch the tool_use block, run your function, and send back a tool_result — in a minimal Python loop.
- 15.03 [Get Reliable JSON Out of the Claude API](https://guides-ai.com/guides/claude-api-structured-output/): Constrain replies with output_config.format, or use messages.parse with Pydantic and Zod, so the Claude API returns JSON that always validates.
- 15.04 [Prompt Caching in the Claude API with cache_control](https://guides-ai.com/guides/claude-api-prompt-caching/): Cache a long system prompt or document with cache_control, confirm hits in the usage fields, and know when the 1.25x write cost pays for itself.
- 15.05 [Claude Message Batches API: Bulk Jobs at Half Price](https://guides-ai.com/guides/claude-api-batch-requests/): Create a batch of Messages requests, poll processing_status, stream results by custom_id — and pay 50% less than synchronous calls.
- 15.06 [Your First OpenAI API Request (Responses API)](https://guides-ai.com/guides/openai-api-first-request/): Create a key, set OPENAI_API_KEY, and make a working Responses API call in curl, Python, and Node — the API OpenAI recommends for new projects.
- 15.07 [OpenAI Structured Outputs: JSON Schema](https://guides-ai.com/guides/openai-api-structured-outputs/): Constrain Responses API output with text.format json_schema, parse it straight into Pydantic or Zod, and reuse strict schemas for function calling.
- 15.08 [Stream LLM Responses: Claude and OpenAI](https://guides-ai.com/guides/stream-llm-responses/): Print tokens as they arrive from the Claude and OpenAI APIs — Python and JavaScript, plus the raw SSE events and how to detect the end.
- 15.09 [Claude Agent SDK for Python: Your First Agent](https://guides-ai.com/guides/claude-agent-sdk-python-quickstart/): Install claude-agent-sdk, run an agent that reads and edits real files, control which tools it may use, and read the final result.
- 15.10 [Claude Agent SDK for TypeScript: Your First Agent](https://guides-ai.com/guides/claude-agent-sdk-typescript-quickstart/): Install @anthropic-ai/claude-agent-sdk, stream an agent that edits real files, gate its tools, and read the final result message.
- 15.11 [Send a PDF to the Claude API (With Page Citations)](https://guides-ai.com/guides/claude-api-pdf-documents/): The document content block in Python, the page and size limits from the docs, citations that point at a page — and when to pre-extract text instead.
- 15.12 [Claude API Thinking and Effort](https://guides-ai.com/guides/claude-api-thinking-and-effort/): Turn on adaptive thinking, read thinking blocks, set depth with output_config.effort (low to max), stream deltas, and know what budget_tokens does.
- 15.13 [Send an Image to the Claude API (base64, URL, Files)](https://guides-ai.com/guides/claude-api-vision-images/): The image content block in Python and curl, the format and size limits from the docs, and a screenshot-to-CSV extraction that doesn't invent numbers.
- 15.14 [Embeddings Explained, With Code](https://guides-ai.com/guides/embeddings-explained-with-code/): What an embedding vector actually is, how to compute two and compare them in Python, and which model and dimension to pick.
- 15.15 [Your First Gemini API Request](https://guides-ai.com/guides/gemini-api-first-request/): Get a Gemini API key from AI Studio, set GEMINI_API_KEY, install the google-genai SDK, and make a working request with the Interactions API.
- 15.16 [Generate Images with the OpenAI API (gpt-image)](https://guides-ai.com/guides/openai-api-image-generation/): Two ways to make an image with the OpenAI API — the Images endpoint and the image_generation tool — save the PNG, set size, quality and format.
- 15.17 [RAG Without a Vector Database: OpenAI File Search](https://guides-ai.com/guides/openai-file-search-rag/): Upload files to an OpenAI vector store, attach the file_search tool to a Responses API call, and get cited answers. No embeddings code, no database.
- 15.18 [A Minimal RAG in Python: Folder In, Cited Answer Out](https://guides-ai.com/guides/rag-with-python-minimal/): Chunk a folder of Markdown, embed it with Voyage, retrieve top-k with a numpy dot product, and answer with Claude citing the chunks it used.
- 15.19 [Upload and reuse files with the Claude Files API](https://guides-ai.com/guides/claude-api-files-api/): Upload a PDF once with the Claude Files API, reference it by file_id across requests in Python, then list and delete it.
- 15.20 [Count tokens and estimate API cost before you send](https://guides-ai.com/guides/count-tokens-and-estimate-cost/): Count tokens locally with tiktoken or Anthropic's count_tokens endpoint, then estimate cost from each provider's published per-token prices.
- 15.21 [Use Gemini API File Search (managed RAG)](https://guides-ai.com/guides/gemini-api-file-search/): Create a managed File Search store with the google-genai SDK, import a file, query it with the file_search tool, and read citations.
- 15.22 [Get structured JSON output from the Gemini API](https://guides-ai.com/guides/gemini-api-structured-output/): Make the Gemini API return JSON that matches a Pydantic schema with the google-genai SDK, then parse it into a typed object.
- 15.23 [Call functions (tools) with the OpenAI API](https://guides-ai.com/guides/openai-api-function-calling/): Give the model your own functions with the OpenAI Responses API: define a tool, run the call it returns, and send the result back in a loop.

### 16 Hugging Face

- 16.01 [Download Models from Hugging Face with the hf CLI](https://guides-ai.com/guides/hugging-face-cli-download-models/): Install the hf CLI, log in, download one GGUF instead of a whole repo with --include, put files where you want with --local-dir, and clean the cache.
- 16.02 [How to Find the Right Model on Hugging Face](https://guides-ai.com/guides/hugging-face-find-the-right-model/): Filter the Models page by task, library, language, license and size, read the sort signals and the model card, pick a file, and search from the API.
- 16.03 [Where the hf CLI downloads models and how to move it](https://guides-ai.com/guides/hugging-face-cache-location/): Find the hf CLI cache path on Windows, macOS and Linux, inspect it with hf cache ls, move it with HF_HOME, and prune old models.
- 16.04 [Access gated models (Llama, FLUX) and fix 401 errors](https://guides-ai.com/guides/hugging-face-gated-models-access/): Accept the model licence, create a read or fine-grained token, log in with hf auth login, set HF_TOKEN for CI, and fix 401/403 errors.
- 16.05 [Run any GGUF from Hugging Face with Ollama or llama.cpp](https://guides-ai.com/guides/run-hugging-face-gguf-with-ollama-llama-cpp/): Pick a GGUF quant in the Files tab, then run it with ollama run hf.co/user/repo or llama-cli -hf user/repo:QUANT.
- 16.06 [Transformers pipeline(): Run a Hugging Face Model Locally](https://guides-ai.com/guides/transformers-pipeline-quickstart/): Install PyTorch and transformers, chat with a small instruct model via pipeline(), set device_map, dtype and max_new_tokens, then classify an image.
- 16.07 [Hugging Face Inference Providers: First API Request](https://guides-ai.com/guides/hugging-face-inference-providers-first-request/): One HF token, one endpoint: call an open model with InferenceClient, the OpenAI SDK or curl, pick a provider, and see how billing works.
- 16.08 [Load, Stream and Filter a Hugging Face Dataset in Python](https://guides-ai.com/guides/hugging-face-datasets-load-stream-filter/): Load a Hub dataset split, inspect its features, stream one too big to download, filter and batch-map it, then save it as Parquet or JSONL.
- 16.09 [Upload a Model or Dataset with the hf CLI](https://guides-ai.com/guides/hugging-face-upload-model-or-dataset/): Log in, create a repo with hf repos create, push files with hf upload and include/exclude, write a minimal README card, and do the same from Python.
- 16.10 [Deploy a Gradio App to Hugging Face Spaces](https://guides-ai.com/guides/deploy-gradio-app-hugging-face-spaces/): Write a minimal Gradio app.py, add requirements.txt and a Space README, create the Space, push with the hf CLI, set secrets, and read the build logs.
- 16.11 [Connect the Hugging Face MCP Server to Claude Code or Cursor](https://guides-ai.com/guides/hugging-face-mcp-server-claude-code/): Add https://huggingface.co/mcp with a read token so your agent searches models, datasets, papers and docs from the chat, and uses MCP Spaces as tools.
- 16.12 [Fix stuck or incomplete Hugging Face downloads](https://guides-ai.com/guides/fix-huggingface-download-incomplete/): Resume a stuck hf download, clear broken .incomplete partials with hf cache prune, and speed transfers with hf_xet when disk space allows.

## Paths (curated reading order)

### Get running
Install, run your first session, give the agent memory, then learn to prompt and pick a model.

1. 01.01 [Install Claude Code on Windows: Download, Setup, Run](https://guides-ai.com/guides/install-claude-code-windows/)
2. 01.02 [How to Install Claude Code on macOS & Linux](https://guides-ai.com/guides/install-claude-code-macos-linux/)
3. 02.01 [Your First Claude Code Session in 5 Minutes](https://guides-ai.com/guides/first-claude-code-session/)
4. 03.01 [How to Set Up a CLAUDE.md That Actually Helps](https://guides-ai.com/guides/create-claude-md-project-memory/)
5. 10.01 [How to Write Better AI Prompts (A Framework)](https://guides-ai.com/guides/write-better-ai-prompts/)
6. 11.01 [Which AI Model Should You Use?](https://guides-ai.com/guides/pick-an-ai-model/)

### Make Claude Code yours
Memory, slash commands, hooks, subagents, permissions, and keeping the bill in check.

1. 03.01 [How to Set Up a CLAUDE.md That Actually Helps](https://guides-ai.com/guides/create-claude-md-project-memory/)
2. 03.02 [Create a Custom Slash Command in Claude Code](https://guides-ai.com/guides/create-claude-code-slash-command/)
3. 03.03 [How to Configure Hooks in Claude Code](https://guides-ai.com/guides/claude-code-hooks/)
4. 03.04 [How to Use Subagents in Claude Code](https://guides-ai.com/guides/claude-code-subagents/)
5. 03.07 [Control What Claude Code Can Do: Permissions](https://guides-ai.com/guides/claude-code-permissions/)
6. 03.06 [How to Track and Cut Claude Code Token Usage](https://guides-ai.com/guides/claude-code-track-costs/)

### Run models locally
Ollama on Windows, LM Studio, the right quantization, the local API, and your own Modelfile.

1. 09.01 [How to Run Ollama on Windows (Install, GPU, Models)](https://guides-ai.com/guides/run-local-llm-ollama-windows/)
2. 09.05 [Run a Local LLM with LM Studio (No Terminal)](https://guides-ai.com/guides/run-llm-lm-studio/)
3. 09.04 [Choose a GGUF Quantization: Q4 vs Q5 vs Q8](https://guides-ai.com/guides/choose-llm-quantization/)
4. 09.03 [Call Your Local Ollama Model from curl & Python](https://guides-ai.com/guides/call-ollama-api/)
5. 09.02 [Ollama Modelfile: a Custom Model With a System Prompt](https://guides-ai.com/guides/ollama-custom-model-modelfile/)

### Prompt like a pro
Better prompts, a real system prompt, chaining, RAG, and fewer hallucinations.

1. 10.01 [How to Write Better AI Prompts (A Framework)](https://guides-ai.com/guides/write-better-ai-prompts/)
2. 10.02 [How to Write a System Prompt That Steers a Model](https://guides-ai.com/guides/write-a-system-prompt/)
3. 10.04 [How to Chain Prompts to Solve a Big Task](https://guides-ai.com/guides/prompt-chaining/)
4. 10.05 [RAG Explained Simply (and When You Need It)](https://guides-ai.com/guides/rag-explained-simply/)
5. 10.03 [How to Reduce AI Hallucinations With Prompts](https://guides-ai.com/guides/reduce-ai-hallucinations/)

### Build with the Claude API
First request, tool use, reliable JSON, prompt caching, batches, and streaming.

1. 15.01 [Your First Claude API Request: curl, Python, Node](https://guides-ai.com/guides/claude-api-first-request/)
2. 15.02 [Claude API Tool Use](https://guides-ai.com/guides/claude-api-tool-use/)
3. 15.03 [Get Reliable JSON Out of the Claude API](https://guides-ai.com/guides/claude-api-structured-output/)
4. 15.04 [Prompt Caching in the Claude API with cache_control](https://guides-ai.com/guides/claude-api-prompt-caching/)
5. 15.05 [Claude Message Batches API: Bulk Jobs at Half Price](https://guides-ai.com/guides/claude-api-batch-requests/)
6. 15.08 [Stream LLM Responses: Claude and OpenAI](https://guides-ai.com/guides/stream-llm-responses/)

### Serve models yourself
Ollama on macOS/Linux, llama.cpp, a GPU server with vLLM, and a chat UI on top.

1. 09.10 [How to Run a Local LLM with Ollama on macOS/Linux](https://guides-ai.com/guides/run-local-llm-ollama-macos-linux/)
2. 09.06 [llama.cpp Quickstart: Run a GGUF Model](https://guides-ai.com/guides/llama-cpp-quickstart/)
3. 09.11 [How to Serve a Model with vLLM](https://guides-ai.com/guides/vllm-serve-openai-compatible/)
4. 09.09 [How to Run Open WebUI with Ollama (Docker)](https://guides-ai.com/guides/open-webui-with-ollama/)

### Try the other coding agents
One first session each: Codex CLI, Aider, Copilot CLI, Windsurf — then Claude Code in CI.

1. 14.01 [How to Get Started with OpenAI Codex CLI](https://guides-ai.com/guides/codex-cli-first-steps/)
2. 14.02 [How to Start Using Aider](https://guides-ai.com/guides/aider-first-steps/)
3. 14.03 [How to Use GitHub Copilot CLI in Your Terminal](https://guides-ai.com/guides/github-copilot-cli-first-steps/)
4. 14.04 [How to Start with Windsurf (Now Devin Desktop)](https://guides-ai.com/guides/windsurf-first-steps/)
5. 04.05 [How to Run Claude Code Headless in a Script or CI](https://guides-ai.com/guides/claude-code-headless-ci/)
6. 04.04 [How to Set Up the Claude Code GitHub Action](https://guides-ai.com/guides/claude-code-github-actions/)

### Hugging Face from zero
Find a model, download it, get past gated repos, run it with Ollama or transformers, then publish your own repo and a Space.

1. 16.02 [How to Find the Right Model on Hugging Face](https://guides-ai.com/guides/hugging-face-find-the-right-model/)
2. 16.01 [Download Models from Hugging Face with the hf CLI](https://guides-ai.com/guides/hugging-face-cli-download-models/)
3. 16.04 [Access gated models (Llama, FLUX) and fix 401 errors](https://guides-ai.com/guides/hugging-face-gated-models-access/)
4. 16.05 [Run any GGUF from Hugging Face with Ollama or llama.cpp](https://guides-ai.com/guides/run-hugging-face-gguf-with-ollama-llama-cpp/)
5. 16.06 [Transformers pipeline(): Run a Hugging Face Model Locally](https://guides-ai.com/guides/transformers-pipeline-quickstart/)
6. 16.09 [Upload a Model or Dataset with the hf CLI](https://guides-ai.com/guides/hugging-face-upload-model-or-dataset/)
7. 16.10 [Deploy a Gradio App to Hugging Face Spaces](https://guides-ai.com/guides/deploy-gradio-app-hugging-face-spaces/)

## Optional

- [About](https://guides-ai.com/about/)
- [Disclosure](https://guides-ai.com/disclosure/)