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 skills/slbug/claude-ruby-grape-rails/compression-reportnpx skills add slbug/claude-ruby-grape-rails --skill compression-reportgit clone --depth 1 https://github.com/slbug/claude-ruby-grape-railsWrote 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/skills/slbug/claude-ruby-grape-rails/compression-report)<a href="https://agentmods.dev/skills/slbug/claude-ruby-grape-rails/compression-report"><img src="https://agentmods.dev/badge/skills/slbug/claude-ruby-grape-rails/compression-report.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 | $0.00037 | $0.02433 |
| Opus 5 | $0.00018 | $0.01216 |
| Sonnet 5 | $0.00007 | $0.00487 |
| Haiku 4.5 | $0.00004 | $0.00243 |
Grade C, and why
rb:compression-report 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.
skill never does. Do not embed literal `rm` / `rm -rf` strings in How it starts
The opening of the file, as written. The whole thing — 211 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compression Telemetry Report
Drafts an anonymized markdown report from the verify-output compression
telemetry the plugin collects under
${CLAUDE_PLUGIN_DATA}/compression.jsonl and hands it back for the
user to file as a GitHub issue. The report is maintainer-actionable
signal that helps the plugin maintainers decide whether a future
release should ship a real replacement mechanism.
Iron Laws
- Never include redacted output verbatim from
--redactwithout reviewing it first. The redactor handles common PII shapes (env-var values, file paths, freeform args) but is not an adversarial sanitizer. Read each cited raw_log file before quoting from it. - Never run
compression-statswithout--redactfor issue bodies. The non-redact JSON contains absolute home-dir paths insideraw_logfields and fullcmdstrings. That is local debug output, not public-issue data. - Never auto-create the GitHub issue. Hand the drafted report to the user and let them open the issue manually. Telemetry is the user's data; only the user can choose to publish it.
- Never read more raw logs than needed to explain the worst few
samples. The report is a digest, not a dump. Keep raw-log Reads
scoped to the redacted_stats
weak_samplesandviolation_sampleslists, capped at the most interesting ~5. - Never delete the user's telemetry yourself. The skill ships no
destructive command. After drafting the report, list the exact
cleanup paths the user can remove when they choose to clean up.
The user composes and runs the deletion command themselves; the
skill never does. Do not embed literal
rm/rm -rfstrings in the drafted output — that mirrors the SessionStart advisory's safer-context posture. - Never quote reconstructed absolute paths in the drafted report
body.
${CLAUDE_PLUGIN_DATA}substitutes inline to a path that contains the user's home directory and plugin install id. Use the resolved path locally forReadand for the cleanup-paths block at the very end of the workflow. Do NOT paste it into the per-sample explanations, citations, or any other prose the user will paste into a shared report.
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 · 211 lines · 37 tokens per session scan C 627e469bba86
rb:compression-report is a skill published in the GitHub repository slbug/claude-ruby-grape-rails (7 stars, last pushed 4d ago), licensed MIT. It adds 37 tokens to every session and 2,433 once invoked, about $0.0002 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 skills, from other repositories
pinme-email
Use this skill when a PinMe project (Worker TypeScript) needs to integrate email sending (sendemail). Guides AI to generate correct Worker TS code.
goai
GoAI is a Go SDK for AI applications. One unified API across 25+ LLM providers. Inspired by the Vercel AI SDK, adapted to Go idioms (generics, interfaces, channels).
verify
Self-healing verification loop (test → clippy → fmt).
rust-project-validate
运行 cargo fmt/clippy/build/e2e/test 验证项目规范；在每次任务结束前必须调用，重要的是必须在端到端测试之后执行，为提交代码做最后准备.
rust-review
Rust 服务审查：panic、SQL 注入、密钥、错误吞没、遗留标记.
python-services
Python patterns for CLI tools, async concurrency, and backend services. Use when working with Python code, building CLI apps, FastAPI services, async with asyncio, background jobs, or configuring uv, ruff, ty, pytest, or pyproject.toml.