Claude Code Mods

Ten mods for Claude Code, built in one night by OneWave AI.

Read the write-up: Claude Code Mods: What They Are, and the Ten We Open-Sourced.

A mod is a Claude Code plugin made of function hooks. It can draw a live pane beside the transcript, a band above the prompt, a status line, or a toast. It can also block, rewrite, or react to any tool call, add slash commands, play sounds, and call the model. Mods hot-reload while you work.

These are a mix of useful and ridiculous. All of them are MIT licensed. Fork them, break them, ship your own.

Mod	Code	Command	What it does
burn-meter	burn-meter/	/burn	Live session cost above the prompt, with plan-limit bars and real-world comparisons
launch-codes	launch-codes/	/launch-codes	Dangerous Bash commands (rm -rf, force push, DROP, curl to bash, prod deploys) need a code before they run
session-wrapped	session-wrapped/	/wrapped	Spotify Wrapped for a session: animated reveal plus a shareable PNG card
boss-fight	boss-fight/	/boss	Failing tests spawn a pixel boss. Each run that fixes tests lands a hit
code-pet	code-pet/	/pet	A pixel pet that eats on tool calls, gets sick on failures, and panics on rm -rf
inner-monologue	inner-monologue/	/monologue	A pane of Claude's dry inner thoughts about your session
sportscaster	sportscaster/	/caster	TV play-by-play of your session, spoken aloud, with crowd effects
agent-narrator	agent-narrator/	/narrate	Every agent step in plain English, with a time-saved counter. Built for showing non-engineers
agent-race	agent-race/	/race	Race several Claude Code sessions on the same task on a live scoreboard
inbox-alerts	inbox-alerts/	/alerts	Gmail, Slack, and Calendar alerts inside Claude Code as toasts, a status count, and a pane

Install

Requires Claude Code 2.1.287 or later.

The fastest way is the plugin marketplace. In a Claude Code session:

/plugin marketplace add OneWave-AI/claude-code-mods
/plugin install burn-meter@claude-code-mods

Swap burn-meter for any mod name in the table. To hack on them instead, clone the repo:

git clone https://github.com/OneWave-AI/claude-code-mods.git ~/claude-code-mods

Try one for a single session:

claude --plugin-dir ~/claude-code-mods/burn-meter

Repeat --plugin-dir to load several. To load mods in every session (including the desktop app), add them to the env block of ~/.claude/settings.json, separated by ::

{
  "env": {
    "CLAUDE_CODE_PLUGIN_DIRS": "~/claude-code-mods/burn-meter:~/claude-code-mods/session-wrapped"
  }
}

Every mod has tests:

claude plugin validate ~/claude-code-mods/burn-meter
claude plugin test ~/claude-code-mods/burn-meter

Most mods have a demo subcommand (/boss demo, /pet demo, /burn demo) so you can see them without waiting for the real event.

The mods

burn-meter

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/burn-meter

A band above the prompt with a growing fire bar for session spend, 5-hour and weekly plan-limit bars with reset times, and the cost converted into burritos and McDoubles. /burn opens the full panel with per-turn cost. Alerts fire when you cross a threshold.

launch-codes

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/launch-codes

Hooks tool.call on Bash and classifies the command before it runs: risky rm -rf, git push --force, git reset --hard, DROP/TRUNCATE through a live SQL client, supabase db reset, vercel --prod, chmod -R 777, and downloaded scripts piped into a shell. A match opens a red-alert pane with a siren and a one-time code. Type the code to arm, press LAUNCH to fire. Anything else is denied and Claude is told why. /launch-codes test <command> shows the verdict without running anything.

It is strict. It blocked one of our own cleanup commands while we were putting this repo together, which is the point.

session-wrapped

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/session-wrapped

/wrapped plays an animated stat reveal (session length, tool calls, MVP tool, red-to-green test runs, longest turn, cost) and writes a 1200px PNG card to your Desktop. Week and month totals come from scripts/usage.py, which reads your local transcripts in ~/.claude/projects and caches the rollups. Needs python3. Nothing leaves your machine.

boss-fight

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/boss-fight

When a test run fails, a pixel boss spawns with one HP per failing test. Every later run that fixes tests lands a hit. Zero failures is a KO with loot. Reads vitest, jest, pytest, mocha, and cargo test output.

code-pet

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/code-pet

A pixel pet in a pane. It eats when Claude calls tools, gets sick when they fail, panics when a destructive command shows up, sleeps when the session is idle, and evolves as you ship. /pet rename <name>, /pet snack.

inner-monologue

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/inner-monologue

While the pane is open, a model call every so often turns the recent prompts and tool calls into one dry line, typed out live. Harmless and weirdly useful for noticing when the agent is flailing.

sportscaster

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/sportscaster

Play-by-play commentary on the session, spoken aloud with generated crowd audio (cheers on passing tests, groans on fouls). /caster booth opens the booth pane. Loud. You have been warned.

agent-narrator

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/agent-narrator

Translates every tool call into one plain-English sentence ("Reading the pricing page to find the old numbers") and keeps a running estimate of time saved. /narrate smart uses a model call per step; the default is rule-based and free. /narrate demo plays a scripted session. We built this for training sessions where the audience has never seen an agent work.

agent-race

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/agent-race

/race start <race> [name] in two or more sessions puts them on the same track. The pane shows each session's tool calls, files touched, and test runs live, and /race done crosses the finish line. Good for comparing models or prompts on the same task.

inbox-alerts

Source · hooks/register.tsx · claude --plugin-dir ~/claude-code-mods/inbox-alerts

Polls Gmail, Slack, and Google Calendar every two minutes through the claude.ai connectors, shows new items as toasts and a count in the status line, and keeps a tabbed Alerts pane. /alerts triage hands the backlog to Claude.

Needs the Gmail, Slack, and Google Calendar connectors connected in claude.ai. Set your email and Slack member ID in /config (or pluginConfigs in settings) so your own messages do not alert you.

Security

A mod is code that runs inside Claude Code with your permissions. It is not sandboxed. Read a mod before you load it, and check what it touches without running it:

claude plugin validate ~/claude-code-mods/<mod>

The hooks: and calls: lines list every event it handles and everything it asks Claude Code to do. What these ten reach:

Mod	Network	Runs processes	Files	Calls a model	Sends data anywhere
burn-meter	No	No	No	No	No
launch-codes	No	No	No	No	No
session-wrapped	No	python3 (reads local transcripts), base64 (writes the PNG)	Writes one PNG to ~/Desktop, a cache in ~/.cache/session-wrapped	No	No
boss-fight	No	No	No	No	No
code-pet	No	No	No	No	No
inner-monologue	No	No	No	Yes, a summary of recent tool calls and the first 120 characters of each prompt	Only to your Claude model, and only while the pane is open
sportscaster	No	No	No	Yes, a summary of recent tool calls	Only to your Claude model
agent-narrator	No	No	No	Only with /narrate smart: the tool name and short fields (path, command), never file contents	Only to your Claude model
agent-race	No	No	Reads and writes ~/.claude/agent-race/<race>/	No	No
inbox-alerts	Through your claude.ai Gmail, Slack and Calendar connectors	No	No	/alerts triage sends alert snippets to Claude	Only to your Claude model

Notes:

• launch-codes is a speed bump, not a security boundary. It checks Bash commands only. It will not stop a determined agent that writes a script file and runs it another way. Keep your normal permissions and sandboxing on.
• inbox-alerts handles text other people wrote. /alerts triage wraps every email and Slack snippet in an <untrusted-alerts> block, strips any attempt to close that block early, and tells Claude to treat the contents as data and never send, reply, or delete anything. That lowers prompt-injection risk; it does not remove it. Review what Claude proposes before acting on it.
• The model-calling mods send tool names, file paths, and short command text to the same Claude model your session already uses. If your commands carry secrets inline, those go too. Keep secrets in environment variables.
• No mod makes its own network requests, and none phones home.

Found a security issue? See SECURITY.md.

Writing your own

Ask Claude Code to make one. The built-in plugin-authoring skill knows the API: say "make a mod that..." and it writes the folder, validates it, and hot-reloads it into the session. Each mod here is three files at minimum:

my-mod/
  .claude-plugin/plugin.json   name, version, description
  hooks/hooks.json             { "modules": ["./register.tsx"] }
  hooks/register.tsx           export const register: Register = (on, options) => { ... }

License

MIT. See LICENSE.

Built by OneWave AI, an Anthropic Partner Network firm that trains teams on Claude and builds with it.