Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/slnu21/pawse/claude-mdgit clone --depth 1 https://github.com/slnu21/PawseWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/slnu21/pawse/claude-md)<a href="https://agentmods.dev/instructions/slnu21/pawse/claude-md"><img src="https://agentmods.dev/badge/instructions/slnu21/pawse/claude-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.02202 | $0.02202 |
| Opus 5 | $0.01101 | $0.01101 |
| Sonnet 5 | $0.00440 | $0.00440 |
| Haiku 4.5 | $0.00220 | $0.00220 |
Grade C, and why
Pawse CLAUDE.md scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
'{"tool_name":"Bash","tool_input":{"command":"rm -rf build"}}' | powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\plugin\claude-widget.ps1 -Event PreToolUse How it starts
The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
What this is
pawse (mascot: the lion Leo) — an interactive widget for Claude Code, built as a single PowerShell + WPF script wired into Claude Code hooks. When Claude stops or needs input, a desktop popup appears; the user types a reply in the widget and Claude continues. Windows-only, no install/build step. See README.md for user-facing docs.
Commands
There is no build/compile — it's an interpreted PowerShell script. Use Windows PowerShell 5.1 (powershell.exe), not pwsh (see STA note below).
# Logic tests (no GUI): danger regex, UTF-8 round-trip, transcript preview, syntax check
powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\plugin\test-widget.ps1
# Manual GUI test — pipe a sample hook JSON to one event
'{"hook_event_name":"Stop","stop_hook_active":false,"transcript_path":""}' | powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\plugin\claude-widget.ps1 -Event Stop
'{"tool_name":"Bash","tool_input":{"command":"rm -rf build"}}' | powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\plugin\claude-widget.ps1 -Event PreToolUse
'{"notification_type":"idle_prompt","message":"hi"}' | powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\plugin\claude-widget.ps1 -Event Notification
# Syntax check without running
$e=$null; [System.Management.Automation.Language.Parser]::ParseFile((Resolve-Path .\claude-widget.ps1),[ref]$null,[ref]$e); $e
Because the widget windows are modal (ShowDialog), automated render-testing launches the script as a child process, sleeps ~2s, checks HasExited (still running ⇒ XAML parsed & window shown), then Kill()s it. See the test harness pattern used during development.
Architecture
claude-widget.ps1 is a single entry point: -Event Stop|Notification|PreToolUse. Flow:
- Read hook JSON from stdin → load
config.psd1→ compute light/dark palette (fromMode, or the system theme viaHKCU:\...\Themes\Personalize\AppsUseLightTheme). switch ($Event)builds an event-specific WPF window from an inline XAML here-string and shows it.- The chosen decision is written to stdout as compact JSON; everything else (the window, errors) must stay off stdout.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 65 lines · 2,202 tokens per session scan C 961f56b47ca4
Pawse CLAUDE.md is an instructions file published in the GitHub repository slnu21/Pawse (1 stars, last pushed 3mo ago), licensed MIT. It adds 2,202 tokens to every session, about $0.0110 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
codex-multi-profile AGENTS.md
Instructions for Hung2124/codex-multi-profile: This repository is a Windows-only Codex Desktop multi-account helper.
cc-safe-setup CLAUDE.md
Instructions for yurukusa/cc-safe-setup, covering project rules, safety, code style and git.
Throughline AGENTS.md
Instructions for kitepon/Throughline, covering agents.md, 最初に読むもの, 製品の所有境界, claude 正本を守る and codex 作業方針.
Throughline CLAUDE.md
Instructions for kitepon/Throughline, covering claude.md, プロジェクト概要, 必読ドキュメント, 実装済みファイルの役割 and コア.
SessionDeck CLAUDE.md
Claude Code instructions for eyalBPM/SessionDeck, covering sessiondeck — working notes for claude code, build, run, deploy, tests, versioning and git.
dotfiles copilot-instructions.md
Copilot instructions for victorfrye/dotfiles: See AGENTS.md for full context, architecture, commands, and conventions.