RTK Integration

Integrate RTK (Rust Token Killer) to reduce LLM token consumption by 60-90% on shell commands. Use when setting up RTK for the first time, checking token sav...

Security Scan

The skill's functionality (installing and using an output-filtering wrapper) is coherent with its description, but the install instructions download and execute a remote script (curl | sh), modify shell profiles, and the package metadata lacks a homepage — these are proportionate but risky and deserve review before running.

Purpose & Capability

The skill's declared purpose (install and use an RTK wrapper to compress shell output) matches the files and instructions: it ships a setup script, docs mapping raw commands to rtk-prefixed equivalents, and guidance to prefer rtk. Nothing requested is unrelated to installing or using an output-filtering CLI.

Instruction Scope

SKILL.md instructs the agent to run the included setup script immediately on install and to always prefer rtk-prefixed commands. The guide also encourages automatic use across many command classes (git, docker, tests) which will suppress raw outputs by design. The install instruction (curl ... | sh) is the main scope concern because it grants broad runtime discretion (execute remote code) and the skill will add PATH lines to user profile files.

Install Mechanism

There is no formal install spec; the provided script uses curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh which executes a remote script directly. Pulling and piping a branch's raw installer is higher risk (content can change) compared with pinned releases or package-managed installs. The script also writes PATH exports into user dotfiles, which is persistent system modification.

Credentials

The skill requests no credentials or environment variables. The only system changes are adding rtk to PATH and running the installer; those are consistent with installing a CLI. No unrelated secrets or config paths are requested.

Persistence & Privilege

The skill does not request always:true and does not request agent-wide privileges, but the setup script persists a PATH export to the user's shell profile (modifies dotfiles) and installs a binary. This is expected for a CLI integration but is a persistent change the user should approve.

What to consider before installing

This skill appears to do what it says, but exercise caution before installing: 1) Inspect the remote installer before running it — avoid blindly piping curl to sh. Open https://github.com/rtk-ai/rtk and review the install.sh content and repository ownership. 2) Prefer installing via trusted package managers (Homebrew, pinned GitHub release, or cargo) rather than the branch raw curl. 3) Back up your shell profile (~/.bashrc, ~/.zshrc) before allowing the script to append PATH lines. 4) Consider testing in a sandbox or VM first; don't install on sensitive production machines until you verify the upstream project. 5) Be aware that always using RTK will suppress raw command outputs by design — keep the ability to run raw commands when full output is needed for debugging or security checks. If you cannot verify the upstream repository or installer, treat this as high-risk and do not run the installer.

RTK filters and compresses shell command outputs before they reach the LLM context window. Works with any agent/model (Claude, MiniMax, GPT, Gemini, OpenClaw, etc.) — model-agnostic.

Setup

1. Install RTK (auto — run on skill install)

When this skill is installed, immediately run the setup script to install RTK if not already present:

bash skills/rtk-integration/scripts/rtk-setup.sh

The script handles installation automatically (Linux/macOS), adds RTK to PATH, and verifies the install.

Manual install if needed:

# Linux/macOS
curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc && source ~/.bashrc

# macOS Homebrew
brew install rtk

# Verify
rtk --version && rtk gain

2. Initialize hooks (optional, for Claude Code / OpenCode auto-rewriting)

rtk init --global          # Claude Code hook
rtk init --global --opencode  # OpenCode hook

⚠️ For OpenClaw agents: hooks don't auto-apply. Use rtk <cmd> explicitly (see below).

3. Check token savings

rtk gain

For config customization, see references/config.md.

---

Core Rule: Always Use RTK Prefixes

When executing shell commands, always prefer the RTK version over the raw command. RTK adds <10ms overhead and reduces output by 60–90%.

See the full command reference: references/commands.md

Quick reference (most common)

# Files
rtk ls .
rtk read <file>
rtk grep "<pattern>" .
rtk find "*.ts" .
rtk diff file1 file2

# Git (most impactful — use on EVERY git operation)
rtk git status
rtk git diff
rtk git log -n 10
rtk git add .          # outputs just "ok"
rtk git commit -m "msg"  # outputs "ok <hash>"
rtk git push           # outputs "ok main"
rtk git pull

# Tests — show failures only (-90%)
rtk test npm test
rtk test npx jest
rtk vitest run
rtk playwright test
rtk pytest
rtk go test
rtk cargo test

# Build & lint
rtk tsc
rtk next build
rtk lint
rtk lint biome
rtk cargo build
rtk cargo clippy
rtk ruff check .

# Docker
rtk docker ps
rtk docker logs <container>

# Package managers
rtk pnpm install
rtk npm install
rtk yarn install

When NOT to use RTK

• Interactive commands (ssh, vim, htop) — use directly
• Already-compressed output (jq-piped JSON, single-line results)
• When you need the full raw output for debugging

---

rtk gain — Token Savings Report

Run rtk gain whenever user asks about token savings. Output example:

📊 RTK Token Savings
════════════════════════════════
Total commands:    1,247
Input tokens:      4.8M
Output tokens:     0.6M
Tokens saved:      4.2M (87.5%)

By Command:
rtk git status    215    1.4M    80.8%
rtk grep          227    786K    49.5%
rtk find          324    6.8M    78.3%