Claude · Workflows

Claude Code Mods: the 3 mods to install first, without writing any code

Three sessions working as a team, a mod that sends simple tasks to the cheapest model, a dashboard inside Claude Code: the copy-paste prompts, plus Anthropic's sample mods and the safety rules.

QQuentin Megevand
November 4, 2026 · 10 min read

Claude Code now has mods: small add-ons that change what Claude does and what you see on screen. You don't write a single line of code: you describe the mod, Claude builds it, you approve it.

This guide gives you the three mods from the video as copy-paste prompts, then more to add, and what to know before you install one. Commands and prices checked on November 4, 2026, on Claude Code 2.1.318 and the official documentation.

Claude AI Lab

The Claude AI Lab is my Skool community where I share my Claude systems and the more advanced modules. Access is $67/month.

Join the Lab →
3
copy-paste mods
2.1.287
minimum Claude Code version
x100
price gap between Haiku 5.5 and Fable 5.1
✅What you need
1
Claude Code 2.1.287 or later. Check with claude --version, update with claude update. In the Code tab of the Desktop app, type /status: you need 2.1.286 or later.
2
A Claude Code session. In the terminal, the Code tab of the Desktop app, or the VS Code extension (2.1.299 minimum for the display). Mods don't run in the regular Claude chat.
3
Nothing to turn on. Mods are on by default.
⚙️How a prompt becomes a mod
4
Paste the prompt into a session. Claude writes the mod in the session's mods folder, using its built-in plugin-authoring skill.
5
Approve each file. In the default mode, Claude Code asks you file by file, because that folder is protected.
6
Pick "Enable for this session". That's hot reloading: the mod loads at the end of the turn, then after every change.
7
Check with /plugin. Press Tab until you reach the Installed tab. To change something, just ask Claude.
1

Sessions that work as a team

👥 3 sessions, 1 shared folder
🧠The idea
8
Your sessions can already talk to each other. One sends a message to another, which reads it between two actions, or starts a turn if it was idle.
9
The mod adds the manager's desk. A panel with every message from the team, and a /standup command that asks each one for a one-line update.
🛠️Set up the team
10
Open 3 sessions in the same folder. A message only carries text, never a file: the shared folder is how they hand work over.
11
Name them. /rename prospector in the first, /rename writer in the second, /rename manager in the third.
12
Paste the right prompt into each session, then the mod prompt into the manager session.
🔎Prospector session
You handle prospecting. List 30 [REAL ESTATE AGENCIES IN DALLAS] that [HAVE NO ONLINE APPOINTMENT BOOKING].
For each one: name, website, public email, and the specific reason [MY OFFER] matters to them.
Save everything to prospects.csv.
When you're done, tell @writer how many prospects you found.
✍️Writer session
You write the outreach emails. As soon as @prospector tells you prospects.csv is ready, read it.
Write one short, personal email per prospect, one file per email, in a drafts/ folder.
Don't send anything.
When everything is written, tell @manager how many drafts are ready.
🧭Manager session
You manage this team: @prospector finds the prospects, @writer writes the emails.
When I ask for an update, ask each of them for one line, then sum it up in three lines: done, blocked, next step.
🧩The team-room mod, to paste into the manager session
Make a mod called team-room.
1. A /team command that opens a panel titled "Team". It shows every message this session sends to my other sessions or receives from them: the time, who, the first line. Newest on top.
2. A /standup command that asks @prospector and @writer, in one line, what they finished and what they're doing next. Each reply appears in the panel as soon as it arrives.
3. Keep the panel's history when the mod reloads.
Write it for my version of Claude Code, then run claude plugin validate and fix whatever comes up.
What a message can't do
It approves nothing on your behalf, runs no command, and permission requests keep coming to you. A session that bypasses permissions holds messages from regular sessions until you agree. Every message received counts toward your usage like a prompt, including those from /standup.
2

The mod that cuts your bill

💸 Haiku 5.5 for simple tasks
🧠The idea
13
By default, every request goes to the same model. You pay big-model prices to fix a typo.
14
The mod sorts each prompt. Simple requests go to Haiku 5.5, the rest stay on your model, and a line above the prompt shows what you saved.
🧮Same task, 20,000 tokens in and 2,000 out
🟢
Haiku 5.5: $0.003
$0.10 input and $0.50 output, per million tokens.
🔵
Sonnet 5.5: $0.06
$2 input and $10 output, per million tokens.
🟠
Opus 5.5: $0.12
$4 input and $20 output, per million tokens.
🔴
Fable 5.1: $0.30
$10 input and $50 output, per million tokens.
🧩The saver mod
Make a mod called saver that sends my simple requests to the cheapest model.
1. On each prompt, classify the request. Simple: fixing a typo, renaming, formatting, a short summary, a quick question. Complex: a plan, debugging, new code, several steps. When in doubt, it's complex.
2. For a simple request, send every request in that turn to the model claude-haiku-5-5, except subagents' requests, and only if my prompt is under 100,000 tokens. Everything else stays on my current model.
3. Above the prompt, show one line: the last model used, and an estimate of what I saved today, based on each request's tokens and list prices (Haiku 5.5: $0.10 input, $0.50 output, per million).
4. Add /saver on and /saver off.
Write it for my version of Claude Code, then run claude plugin validate and fix whatever comes up.
Why 100,000 tokens
Above that, Haiku 5.5 goes up to $0.50 input and $2.50 output, five times its normal price.
The cache trap
Each model has its own cache: after a switch, the next request reads the whole conversation with no cache hits. So the mod pays off most on short sessions. On a long session with a big context, turn it off with /saver off. On a subscription, the amount shown is still an estimate: your real measure is /usage.
3

Your dashboard inside Claude Code

📊 /hq command
🧠The idea
15
A mod can draw. A panel beside the conversation in a wide fullscreen terminal, above the prompt otherwise, with tabs, buttons, text fields and even charts.
16
Connect Gmail for the To answer tab. Check with /mcp that the connector is there. Without it, the tab says so in one line and the rest works.
17
Keep /hq as the command. Claude Code reserves the name /dashboard and refuses to register it for a mod (checked with the validator, version 2.1.318).
🧩The dashboard mod
Make a mod called dashboard.
1. A /hq command that opens a panel with three tabs:
- To do: a field to add a task, a button to tick off each task, a progress chart. The list carries over between sessions.
- To answer: emails from the last 24 hours I haven't replied to, read from my Gmail connector. One row per email: who, subject, how long ago. If there's no Gmail connector, say so in one line.
- Limits: my 5-hour and weekly usage limits, as two bars.
2. Above the prompt, a single line, for example: 3 to do · 2 to answer · 40% of my 5-hour limit.
3. Refresh the To answer tab every 10 minutes.
Write it for my version of Claude Code, then run claude plugin validate and fix whatever comes up.
📈Next
18
Add a tab for each number you check every morning. Today's leads, sales, ad spend: a mod can call a tool connected through MCP or query a web address. Ask Claude, and the mod reloads at the end of the turn.
4

More mods to add

🧩 3 official, 4 in one sentence
🏷️The 3 mods published by Anthropic
🌦️
token-weather
A forecast of your context window, above the prompt.
💥
blast-radius
Holds a risky command, such as rm -rf or a force push, shows what it would change, then lets you proceed or cancel.
🎬
replay-theater
A /replay command that steps through the file edits from the last turn.
▶️Try blast-radius for one session
git clone https://github.com/anthropics/claude-code-playground
claude --plugin-dir claude-code-playground/claude-code/mods/blast-radius
🌿Your git branch, always in view
Make a mod that shows my git branch above the prompt.
🧭Your remaining context
Make a mod that shows, above the prompt, how much context I have left, in red below 20%.
🛑A guardrail on publishing
Make a mod that holds any git push to main and asks me to confirm with two buttons: proceed or cancel.
📝A session wrap-up
Make a mod that adds a /wrapup command: three lines on what was done in this session, written by Haiku 5.5 and appended to wrapup.md.
You already have some
/diff is a mod. So is "You should know", a side agent you can turn on, which leaves a note above the prompt when it spots something you might have missed. Look in /plugin, Installed tab, under Built-in.
5

Keep your mods, without the risk

🔒 read this before you install
📦Keep a mod
19
A mod Claude writes only loads in its own session. Its folder even gets deleted along with old sessions (the cleanupPeriodDays setting).
20
Move it out as soon as you like it, with the prompt below, so you get it in every session.
💾The prompt to keep your mods
Copy every mod written in this session into ~/mods.
Turn ~/mods into a local marketplace called my-mods, and install each mod from it so it loads in all my sessions.
Then show me /plugin.
⚠️What a mod can do once loaded
21
It acts as you, with no sandbox. It reads and writes your files, starts programs, and reaches the network.
22
It reads your secrets. Environment variables and settings, API keys included.
23
It sees and changes your session. Every prompt, every tool call, which it can approve without asking you. It can also spend your usage by calling a model.
🛡️Your 3 reflexes
24
Install only from authors you trust. A mod Claude wrote for you is yours. Read anything else first.
25
Run the validator before you install. claude plugin validate ./some-mod runs nothing: the hooks: line lists what the mod listens for, the calls: line what it asks to do.
26
Know the off switches. One mod: /plugin, Installed tab. All of them for one session: claude --safe-mode. All of them everywhere: "disableAllHooks": true in ~/.claude/settings.json.
What no mod can touch
The permission prompt. A mod can restyle most of Claude Code, but never what that prompt shows you.

Where to start

Install one mod tonight. You want to see your costs drop: saver. You juggle several terminals: team-room. You want to open Claude Code like your home screen: dashboard.

To go further, the official mods documentation covers everything they can do, the mods published by Anthropic show how they're built, and the prices above come from the official Claude API pricing page.

Want to go further?

And day-to-day, I post one reel a day on Instagram: @quentin_iamarketing