agentleFS
The-Syntax-SlayerThe-Syntax-Slayer/skill-genie/skills/rtk-token-saver/SKILL.md

rtk-token-saver

Skill1 starsChanged 5 months agoMITView on GitHub

Use rtk to reduce token usage from common shell, git, search, file-reading, test, build, and lint commands while preserving raw-output fallbacks when exact output is required.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

21 rules. About 0 are common, 21 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈21

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when a task involves shell commands that may produce large output, especially repository inspection, git history, diffs, …

    line 10
  2. Distinctive:

    Use RTK by default for:

    line 16
  3. Distinctive:

    Search commands across large repositories.

    line 19
  4. Distinctive:

    Test, lint, typecheck, and build commands where failures are the main signal.

    line 21
  5. Distinctive:

    Do not use RTK when:

    line 24
  6. Distinctive:

    Before relying on RTK, check whether it is available:

    line 33
  7. Distinctive:

    If RTK is missing, do not install it silently.

    line 41
  8. Distinctive:

    Instead of broad raw commands such as:

    line 57
  9. Distinctive:

    Use native grep, ripgrep, or repository search tools when exact match behavior, line ranges, or complete output is required.

    line 74
  10. Distinctive:

    Use raw git commands when applying patches, generating exact diffs for review, or feeding output into another tool.

    line 89
  11. Distinctive:

    Prefer RTK wrappers for high-noise checks:

    line 93
  12. Distinctive:

    If a compact failure summary is not enough to fix the issue, rerun the smallest failing command natively with flags that show the exact …

    line 109
  13. Distinctive:

    Use raw output only when full logs are required for auditing, exact reproduction, or external attachment.

    line 125
  14. Distinctive:

    Check whether RTK is installed when the task would benefit from compact command output.

    line 129
  15. Distinctive:

    Use the RTK command that corresponds to the native command.

    line 130
  16. Distinctive:

    If RTK hides information needed for correctness, rerun the native command with the narrowest scope that exposes the missing detail.

    line 132
  17. Distinctive:

    Never use RTK as an excuse to skip validation.

    line 137
  18. Distinctive:

    Never claim a test, lint, typecheck, or build passed unless the command exit status succeeded.

    line 138
  19. Distinctive:

    Do not use RTK for commands whose output must be consumed exactly by a script or parser.

    line 139
  20. Distinctive:

    Do not run broad commands just because RTK compresses output.

    line 140
  21. Distinctive:

    When RTK was used for important work, summarize:

    line 145

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

RTK Token Saver

Use this skill when a task involves shell commands that may produce large output, especially repository inspection, git history, diffs, search, file reading, tests, linting, builds, logs, containers, or package manager output.

RTK is a CLI proxy that filters and compresses command output before it enters the agent context. Prefer RTK when compact output is enough to make the next decision. Use raw commands when exact bytes, full JSON, full logs, cryptographic material, or complete unmodified output is required.

Activation Rules

Use RTK by default for:

  • Directory listings and tree-like exploration.
  • Git status, diffs, logs, pull, push, and GitHub CLI summaries.
  • Search commands across large repositories.
  • Reading large files when a concise view is sufficient.
  • Test, lint, typecheck, and build commands where failures are the main signal.
  • Docker, Kubernetes, AWS, package manager, and log output that may be noisy.

Do not use RTK when:

  • The user explicitly asks for raw or complete output.
  • The next step requires exact formatting, exact bytes, or complete structured data.
  • A command output will be parsed programmatically by another command.
  • Security-sensitive values may be exposed; avoid reading secrets at all.
  • RTK is not installed or the relevant RTK wrapper fails. Fall back to the native command and keep output scoped.

Install and Verify

Before relying on RTK, check whether it is available:

which rtk
rtk --version
rtk gain

If RTK is missing, do not install it silently. Report that RTK is unavailable and run the native command with the narrowest safe scope.

Command Mapping

Repository and File Exploration

Prefer:

rtk ls .
rtk find "*.ts" .
rtk read path/to/file.ts
rtk read path/to/file.ts -l aggressive
rtk smart path/to/file.ts

Instead of broad raw commands such as:

ls -laR .
find . -type f
cat path/to/large-file.ts

Search

Prefer:

rtk grep "pattern" .
rtk grep "functionName" src

Use native grep, ripgrep, or repository search tools when exact match behavior, line ranges, or complete output is required.

Git

Prefer:

rtk git status
rtk git diff
rtk git diff --stat
rtk git log -n 10
rtk gh pr list
rtk gh pr view 42

Use raw git commands when applying patches, generating exact diffs for review, or feeding output into another tool.

Tests, Lint, Typecheck, and Build

Prefer RTK wrappers for high-noise checks:

rtk test npm test
rtk jest
rtk vitest
rtk pytest
rtk cargo test
rtk go test
rtk lint
rtk tsc
rtk next build
rtk cargo clippy
rtk ruff check

If a compact failure summary is not enough to fix the issue, rerun the smallest failing command natively with flags that show the exact failure.

Logs and Services

Prefer:

rtk docker ps
rtk docker logs <container>
rtk docker compose ps
rtk kubectl pods
rtk kubectl logs <pod>
rtk aws logs get-log-events
rtk log path/to/app.log

Use raw output only when full logs are required for auditing, exact reproduction, or external attachment.

Workflow

  1. Check whether RTK is installed when the task would benefit from compact command output.
  2. Use the RTK command that corresponds to the native command.
  3. Inspect the compact output and decide the next smallest action.
  4. If RTK hides information needed for correctness, rerun the native command with the narrowest scope that exposes the missing detail.
  5. Record validation commands exactly, including whether RTK or native output was used.

Guardrails

  • Never use RTK as an excuse to skip validation.
  • Never claim a test, lint, typecheck, or build passed unless the command exit status succeeded.
  • Do not use RTK for commands whose output must be consumed exactly by a script or parser.
  • Do not run broad commands just because RTK compresses output; keep command scope targeted.
  • Respect project-specific tool preferences. If a project provides a preferred test or lint command, wrap that command with RTK only when compact output is appropriate.

Completion Output

When RTK was used for important work, summarize:

  • The RTK commands run.
  • Whether any native command was rerun for exact output.
  • Final validation status.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.