Tutorials for running AI agents
Practical guides to Claude Code and Codex: how to run them headless, on a schedule and in the background, and how to give them ongoing work. Every post has steps you can run.
OpenClaw Reviews: What Hacker News Users Say (2026)
OpenClaw reviews from 10 Hacker News threads, February to November 2026: what users like, the most common complaints, and what they switch to instead.
Read the guide →Grok Bot Reviews: What Hacker News Users Say (2026)
Grok Bot reviews from 6 Hacker News threads, August to November 2026: what users like, complaints about usage limits and security, and what they use instead.
Read the guide →AI Coworker Tools Compared: Claude Cowork, Copilot, ChatGPT Work
AI coworkers compared: Claude Cowork, Copilot Cowork, ChatGPT Work, Gemini, Meta Muse, Grok Bot and open-source picks like Andrew Ng's OpenWorker.
Read the guide →Gemini 4 Hydrogen: Pricing, Benchmarks, vs Argon and 3.8 Flash
Gemini 4 Hydrogen, Google's fast coding and agent model: $1.50 / $7.50 per million tokens, 1M context, four thinking levels, benchmarks, and how it compares with Argon, Carbon, 3.8 Flash and Claude.
Read the guide →Multi-Agent Harness: The Harness for an AI Agent Team
One agent's harness isn't enough once you run a team of agents. A multi-agent harness adds roles, a shared task system, a report chain, agents that talk to each other and scheduled wake-ups.
Read the guide →Claude Fable 5.2 Videos: 32 Films and 3D Scenes Made in Code
Videos made with Claude Fable 5.2: an anime short rendered in Blender, Istanbul from the air, a golden ship above the clouds, landing pages and science visuals. No video model: Fable writes the code. Each with the maker, likes and prompt.
Read the guide →Claude Fable 5.2 Games: 28 Games You Can Play or Watch
Games made with Claude Fable 5.2: a AAA-looking FPS, a Van Gogh bike ride, a voxel Earth, kart racers and boss fights, most playable in the browser. Each with the maker, play link, likes and, where shared, the prompt.
Read the guide →Opus 5.5 Motion Graphics: 43 Examples and Their Prompts
Motion graphics videos made with Claude Opus 5.5: launch films, logo reveals, kinetic type and explainers, each with the maker's prompt, likes and views. Plus the prompt everyone copied and how to make one in Claude Code.
Read the guide →Claude Fable 5.2: Benchmarks, Pricing, vs Opus 5.5
Claude Fable 5.2, released November 9, 2026: what's new, Anthropic's benchmarks, API pricing ($8/$40) and fast mode, 1M context, Fable 5.2 vs Opus 5.5, vs Fable 5.1 and vs Mythos 5.2, and how to use it in Claude Code.
Read the guide →Aider Alternatives (2026): Is Aider Dead? What to Use Now
Is Aider still maintained? Its last release, last commit and open issues, then the best Aider alternatives in 2026: the cecli fork, OpenCode, Pi, Crush, Cline, Nanocoder, Codex CLI and Claude Code, with install commands.
Read the guide →OpenCode Alternatives (2026): Free, Open Source and Local
The best OpenCode alternatives in 2026: open-source terminal agents (Pi, Crush, Kilo CLI, Goose, Cline, Nanocoder), free options with no API key, Rust and Windows picks, local LLMs, GUIs, and Claude Code, Codex and Copilot CLI.
Read the guide →Devin Alternatives (2026): Open Source, Free and Cheaper
The best Devin AI alternatives in 2026, with current prices from each vendor: open-source agents you can install today (OpenHands, OpenCode, mini-swe-agent, Cline, Goose), cloud agents from OpenAI, Anthropic, Google and GitHub, and team platforms.
Read the guide →Claude Code Channels: Telegram, Discord, iMessage Setup
How Claude Code channels work and how to set up the official Telegram, Discord and iMessage plugins: commands, pairing, sender allowlists, requirements, why channels aren't available, and Slack.
Read the guide →Codex Remote Control: Use Codex From Your Phone
How to control Codex on your Mac or Windows PC from the ChatGPT app: setup with a QR code, SSH hosts, what you can do from the phone, what the host needs, and fixes when remote control isn't working.
Read the guide →Codex Long-Running Tasks: Run Codex Overnight
How to keep OpenAI Codex working for hours or overnight: /goal and /plan, cloud tasks, codex exec resume, keeping the Mac awake, checking in from your phone, usage limits and what stops it.
Read the guide →Can Claude Code Run 24/7? Run It Overnight, Unattended
How to keep Claude Code working overnight or around the clock: permission modes, /goal, keeping the machine awake, cloud sessions, scheduled tasks, usage limits, and what stops it.
Read the guide →Claude Code Agent Teams vs Subagents: Setup | Crewmark AI
Claude Code agent teams explained from the docs: how they differ from subagents, how to enable and start one, tmux and iTerm2 split panes, team size, token costs and limitations, with example prompts.
Read the guide →Hermes Agent vs OpenClaw vs Claude Code (2026) | Crewmark AI
Hermes Agent vs OpenClaw, and how both compare with Claude Code: what each is for, models and subscriptions, memory and skills, security, and how to migrate from OpenClaw to Hermes. From each project's own docs.
Read the guide →OpenClaw Alternatives (2026): Hermes, NanoClaw | Crewmark AI
The OpenClaw alternatives compared from their own docs: Hermes Agent, NanoClaw, ZeroClaw, IronClaw and NVIDIA NemoClaw. How each runs, which models and subscriptions it supports, its security model, and when a work-focused option like Crewmark fits better.
Read the guide →Claude Squad Alternatives (vs Conductor) | Crewmark AI
What Claude Squad does, where it falls short (tmux only, no Windows, you start every task), and the alternatives for running several Claude Code or Codex agents: Claude Code's own worktrees, Paradev, Kanban Code, Conductor and Crewmark.
Read the guide →Claude Code Kanban Board: Tasks and Tools | Crewmark AI
Claude Code has a task checklist (Ctrl+T) but no kanban board. How the task list works, how to put it on a board, and the kanban tools for Claude Code sessions: claude-code-kanban, Kanban Code, Claudine, Vibe Kanban and Crewmark.
Read the guide →Vibe Kanban Alternatives After the Shutdown | Crewmark AI
Vibe Kanban's company shut down and its cloud board is gone. What still works, how to keep your issues, and the alternatives for running Claude Code and Codex from a board: Paradev, Nimbalyst, Claude Squad, Kanban Code and Crewmark.
Read the guide →GitHub Copilot in Slack: Setup, Alternatives | Crewmark AI
@GitHub in Slack and Teams is generally available since November 6, 2026. How to set it up, what the Copilot agent does from a chat thread, permissions and limits, and how it compares with Claude Code in Slack.
Read the guide →AI SEO Agent: Build One With Claude Code | Crewmark AI
What an AI SEO agent can do, the daily workflow that works, and how to build one with Claude Code or Codex: CLAUDE.md rules, skills, MCP, IndexNow and a schedule. Based on the agent that runs this site's SEO.
Read the guide →DeepSeek V4.1 Flash (1017): Price, Benchmarks, vs V4 Pro
DeepSeek-V4.1-Flash-1017 explained: what changed on October 17, 2026, API price, 1M context, open weights on Hugging Face, DeepSeek's benchmark results, and when to use V4 Pro instead.
Read the guide →Claude Code with DeepSeek V4: Setup and Cost | Crewmark AI
Run Claude Code on DeepSeek's models: the environment variables DeepSeek documents for deepseek-flash and deepseek-v4-pro, model mapping, web search, pricing, and the Codex setup.
Read the guide →DeepSeek Harness: What It Is, How to Install (Linux Too)
DeepSeek Harness (dsh) explained: what the open-source agent from DeepSeek does, how to install it on Mac, Windows and Linux (web UI or from GitHub), DeepSeek API pricing, and how it compares with Claude Code and Codex.
Read the guide →What Is a Superagent? DeepSeek, Base44, ClickUp Compared
"Superagent" names several AI products at once. What a superagent is, why DeepSeek's Super-agent is trending, and how it compares with Base44 Superagents, ClickUp Super Agents, Genspark and the open-source superagent-ai projects.
Read the guide →Codex AGENTS.md: File Location and Examples | Crewmark AI
How Codex reads AGENTS.md: the global ~/.codex file, project and nested files, AGENTS.override.md, the load order and 32 KiB limit, fallback filenames, code review rules, and examples.
Read the guide →Best Cursor Alternatives (2026): Free, Open-Source, Cheaper
Cursor joined SpaceX and moved to Grok-first plans. Ten alternatives compared: Claude Code, Codex, GitHub Copilot, Devin Desktop (Windsurf), Zed, Cline, Kilo Code, OpenCode, Antigravity and Junie, with prices and free tiers.
Read the guide →Claude Code vs Cursor (2026): Pricing, Limits | Crewmark AI
Claude Code vs Cursor compared in November 2026: Cursor's new Grok-first plans after joining SpaceX, what Claude models cost in each, agents, CLI, rules, MCP and hooks, and how to choose.
Read the guide →Claude Code MCP: Servers, Config and .mcp.json | Crewmark AI
How to set up MCP servers in Claude Code: claude mcp add for HTTP and stdio servers, local, project and user scopes, where the config file lives, .mcp.json with environment variables, OAuth and troubleshooting.
Read the guide →Claude Code Hooks: Events and Examples | Crewmark AI
Claude Code hooks explained: the lifecycle events, where hooks go, exit codes and JSON output, and copy-paste examples for notifications, auto-formatting, blocking commands and Stop hooks.
Read the guide →Claude Code Token Usage: 12 Tips to Reduce It | Crewmark AI
Why Claude Code uses so many tokens and how to cut it: check usage with /usage and /context, clear and compact, pick the model and effort, trim CLAUDE.md and MCP, use hooks, skills and subagents.
Read the guide →Best Claude Code Orchestration Tools (2026) | Crewmark AI
Run several Claude Code agents at once: the built-in options (subagents, agent view, agent teams, workflows) and 10 orchestration tools compared, from Claude Squad and Orca to Paperclip, Multica and Crewmark.
Read the guide →Claude Inference Hooks: How They Work, Setup | Crewmark AI
Claude inference hooks, generally available since November 5, 2026, send every prompt and tool call to your security server for an allow or deny verdict. How they work, setup, limits, and how they differ from Claude Code hooks.
Read the guide →Best AI Employee Platforms (2026): 8 Compared | Crewmark AI
Lindy, Sintra, Marblism, Artisan, Ema, Devin, the open-source ai-employees project and Crewmark compared: roles, how work starts, which models they run on, free plans and prices.
Read the guide →Claude Code Plan Mode: Shortcut, vs Auto Mode | Crewmark AI
How plan mode works in Claude Code: the Shift+Tab shortcut and /plan, what Claude can do while planning, approving a plan, making it the default, and plan mode vs auto, accept edits and manual.
Read the guide →Claude Code Skills: How to Create, Examples | Crewmark AI
What Claude Code skills are, how they load, how to create one with SKILL.md, examples, supporting files, running a skill in a subagent, installing skills from plugins, and best practices.
Read the guide →Claude Code Slash Commands: Cheat Sheet | Crewmark AI
A cheat sheet of the Claude Code slash commands worth knowing, how to write your own custom slash commands with arguments, and how commands relate to skills.
Read the guide →Conductor Alternatives as Conductor Joins Meta | Crewmark AI
Conductor, the Mac app for parallel Claude Code and Codex agents, is joining Meta. The shutdown dates, how to save your workspaces, and the alternatives for Mac, Windows and Linux.
Read the guide →Claude Code Worktree: Workflow and Cleanup | Crewmark AI
How git worktrees work in Claude Code: claude --worktree, where worktrees go, the base branch, .worktreeinclude for .env files, cleanup, subagent isolation and the settings.
Read the guide →Claude Code vs Codex (2026): Pricing, Limits | Crewmark AI
Claude Code vs OpenAI Codex in November 2026: models, plans and prices, 5-hour and weekly limits, CLAUDE.md vs AGENTS.md, sandboxing, subagents, scheduling, and which to pick.
Read the guide →CLAUDE.md: Best Practices, Template, AGENTS.md | Crewmark AI
How CLAUDE.md works in Claude Code: where the files go and in what order they load, imports, rules, what to put in it, a template, and how it relates to AGENTS.md.
Read the guide →Claude Duet: Waitlist, Access and Limits | Crewmark AI
Claude Duet is Anthropic's proactive agent app, in research preview since October 21, 2026. What it does, who can get it, how to join the waitlist, what it costs, its safety limits, and how it compares.
Read the guide →Claude Message Threads: Faster Agent Sessions | Crewmark AI
Anthropic's message threads send a Claude conversation once, then only what's new. What changes for Claude Code, how to use the beta API, the 5-minute expiry and the limits.
Read the guide →DeepSeek Super-Agent (dsh-superagent-prototype): Setup
The dsh-superagent-prototype repo on GitHub, explained: DeepSeek's open-source super agent that works on one GitHub repo by itself. What it does, what it needs, step-by-step setup and its safety limits.
Read the guide →ChatGPT Dots: How to Use Them and Price | Crewmark AI
ChatGPT dots are OpenAI's always-on agents. What they do, which plans get them, how to set one up, the custom rules, the limits, and how they compare.
Read the guide →Multiple Claude Code Agents: Subagents, Worktrees, Teams
Run several Claude Code agents at once: subagents, parallel sessions in git worktrees, background sessions with --bg, and experimental agent teams, without edit conflicts.
Read the guide →Claude Code Background Agents: Headless Mode | Crewmark AI
Run Claude Code in the background: background tasks, agent view with --bg, headless claude -p with cron, cloud sessions, scheduled tasks and the permissions each needs.
Read the guide →