i-have-adhd vs port22: Features, Pricing & Which Is Better (2026)
A side-by-side comparison of i-have-adhd and port22 — features, pricing, and ideal use cases — to help you decide which AI tool fits your workflow.
i
i-have-adhd
ayghri
Coding-agent skill that forces action-first, numbered answers — kills the 'Great question! Let me think...' preamble in Claude Code, Codex, and friends.
Key features
- Action-First Output Rule: Every response leads with the next concrete action instead of preamble.
- Numbered Multi-Step Tasks: Coding tasks come back as numbered lists capped at 5 items each.
- Preamble & Closer Suppression: Eliminates 'Great question!', 'Hope this helps', and other filler.
- Specific Time Estimates: Forces the agent to answer with minutes rather than vague 'a bit' phrasing.
- One-Command Install: /i-have-adhd in Claude Code, $i-have-adhd in Codex, or persistent activation via ~/.claude/.i-have-adhd.
- Auto-Update in Claude Code: Claude Code fetches the skill from GitHub and keeps it up to date — no local clone needed.
- Multi-Language Skill Docs: README available in English, Simplified Chinese, Japanese, Korean, Vietnamese, and Portuguese (BR).
- Fork-and-Tune: Copy skills/i-have-adhd/SKILL.md, edit the 10 rules to taste, restart the agent, and re-invoke.
Best for
- Reduce Preamble Fatigue: Developers who lose time scanning past 'Great question!' openings get straight-to-the-point agent output.
- ADHD-Friendly Workflow: Neurodivergent developers get responses formatted the way their attention can actually consume.
- Terminal-First Sessions: Users driving Claude Code or Codex from a terminal want short, numbered instructions rather than essays.
- Team Style Standard: Teams adopt the skill so every developer's coding agent responds in the same terse format.
- Custom Response Rules: Fork the SKILL.md and hand-tune the 10 rules to fit an internal engineering culture or accessibility need.
port22
port22
Mobile remote for AI coding agents — pair your phone with a Mac companion and drive Claude Code, Codex, or OpenCode from anywhere.
Key features
- Attach to a Live Session: Drives the agent session you already started in your terminal rather than spawning a fork or shadow instance.
- Multi-Agent Support: Works with Claude Code, Codex, and OpenCode today, with OpenClaw and Hermes on the roadmap.
- Live Token Streaming: Reads every token the agent produces as it thinks, edits files, and runs commands, streamed to your phone in real time.
- Push Approvals: Sends a push the moment an agent finishes or needs input so you can approve or answer without opening the app.
- Dynamic Island & Lock-Screen Status: Shows a live count in the Dynamic Island and the full session roster on the lock screen.
- iCloud Pairing: Same Apple ID on Mac and iPhone auto-pairs devices — no new login and no QR code scanning.
- LAN-First with Encrypted Relay: Uses the local network at your desk and falls back to an end-to-end-encrypted relay when remote, keeping code off port22 servers.
Best for
- Approve Long-Running Agents on the Go: Kick off a Claude Code refactor and approve intermediate steps from your phone during commute or errands.
- Monitor Overnight Runs: Watch an agent's live transcript from bed or a meeting without returning to the desk.
- Cross-Room Development: Continue a coding session from the couch while the actual model, tools, and repo stay on the Mac.
- Multi-Session Oversight: Track several parallel agent sessions from one phone view instead of tab-switching on the Mac.
- Remote Debugging Assist: Read an agent's reasoning stream from a client site or coffee shop before deciding whether to intervene.