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/patrickjaja/claude-desktop-extra/agents-mdgit clone --depth 1 https://github.com/patrickjaja/claude-desktop-extraWrote 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/patrickjaja/claude-desktop-extra/agents-md)<a href="https://agentmods.dev/instructions/patrickjaja/claude-desktop-extra/agents-md"><img src="https://agentmods.dev/badge/instructions/patrickjaja/claude-desktop-extra/agents-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 | $0.08822 | $0.08822 |
| Opus 5 | $0.04411 | $0.04411 |
| Sonnet 5 | $0.01764 | $0.01764 |
| Haiku 4.5 | $0.00882 | $0.00882 |
Grade D, and why
claude-desktop-extra AGENTS.md scanned grade D with 3 findings 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 yesterday.
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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
sudo pacman -U build/claude-desktop-extra-*.pkg.tar.zst Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf ~/.config/Claude/local-agent-mode-sessions/ Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
**Rule when adding a new patch:** if it writes to a fixed user-level path, prefer `app.getPath("userData")` (auto-isolates) over `os.homedir()+"/.config/Claude"` (single-instance leak). If it opens a Unix socket or pipe How it starts
The opening of the file, as written. The whole thing — 488 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md - Project Guidelines
Project Overview
This project repackages Anthropic's official Claude Desktop Linux .deb for the distros Anthropic does not ship (Arch via our own pacman repo primary, plus Fedora/RHEL via RPM, NixOS, AppImage, and our own Debian/Ubuntu .deb). It applies JavaScript patches to the official build's app.asar to add Linux-only value-adds (Computer Use, custom themes, multi-profile, Quick Entry) and Linux fixes.
The official .deb (apt repo https://downloads.claude.ai/claude-desktop/apt) bundles its own Electron (42.10.0 as of v1.37937.0; authoritative source: the deb tree's usr/lib/claude-desktop/version file) and a native Cowork VM backend. We download it, verify GPG + SHA256, dpkg-deb -x it, patch its app.asar, and repackage. We no longer ingest the Windows MSIX, and we no longer pin Electron or rebuild node-pty (both are bundled in the official .deb).
Layout invariant (2026-07-13): every package ships the .deb's usr/lib/claude-desktop/ tree verbatim (binary renamed claude-desktop -> claude; patched resources/app.asar exe-adjacent; our CU bridges added to resources/). Electron auto-loads the adjacent asar - the official build's OnlyLoadAppFromAsar + EnableEmbeddedAsarIntegrityValidation fuses are on (integrity is not enforced on Linux, which is why a patched asar boots). The launcher passes NO asar on the command line, so process.resourcesPath and app.isPackaged behave exactly as on the stock Anthropic .deb. Never add a patch that redirects resource paths; put bundled files at their upstream resources/ position instead. On NixOS the nixpkgs Electron dist is merged with our resources/ into one directory to preserve the same adjacency.
Target platform: Linux only. We do NOT need macOS or Windows compatibility — all patches target Linux exclusively (X11, Wayland, XWayland).
Architectures: x86_64 (primary) and aarch64 (Raspberry Pi 5, NVIDIA Jetson/DGX Spark, etc.). Any native binary we ship (kwin-portal-bridge) must be built for both architectures. (node-pty arrives pre-built for both arches inside the official .deb.)
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.
- yesterday Changed · +1 lines · +249 tokens per session 3aad64d6dddf
- 5d ago First seen · 487 lines · 8,573 tokens per session scan D 58f2e10ed1f9
claude-desktop-extra AGENTS.md is an instructions file published in the GitHub repository patrickjaja/claude-desktop-extra (615 stars, last pushed 2d ago), licensed MIT. It adds 8,822 tokens to every session, about $0.0441 per session on Opus 5. A static security scan graded it D with 3 findings (asks for root, recursive force delete, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
claudebar CLAUDE.md
Claude Code instructions for mryll/claudebar, covering claude.md, non-obvious rules and hardening invariants.
codexbar CLAUDE.md
Claude Code instructions for mryll/codexbar, covering claude.md, tooling and non-obvious rules.
codex-desktop-linux AGENTS.md
AGENTS.md instructions for ilysenko/codex-desktop-linux, covering agents.md, purpose, non-negotiable rules, source routing and editing guidance.
Unity-MCP copilot-instructions.md
Copilot instructions for IvanMurzak/Unity-MCP, covering project guidelines, critical rules, no c# reflection for private access, code review and code style.
mcp-erpnext AGENTS.md
AGENTS.md instructions for Casys-AI/mcp-erpnext, covering agents.md, project overview, project structure & module organization, build, test, and development commands and run all tests (also: deno task test).
llm-context.py CLAUDE.md
Instructions for cyberchitta/llm-context.py, covering claude.md, working notes (gitignored) and draining the field notes.