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 skills add DeepLink-org/probing --skill slow_rankgit clone --depth 1 https://github.com/DeepLink-org/probingWrote 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/deeplink-org/probing/slow_rank)<a href="https://agentmods.dev/skills/deeplink-org/probing/slow_rank"><img src="https://agentmods.dev/badge/skills/deeplink-org/probing/slow_rank.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 5.1 | $0.00013 | $0.00329 |
| Opus 5 | $0.00006 | $0.00164 |
| Sonnet 5 | $0.00003 | $0.00066 |
| Haiku 4.5 | $0.00001 | $0.00033 |
Grade A, and why
slow_rank scanned grade A with 0 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 8d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Find slow rank (straggler)
对比各 rank 的 collective 延迟，找出明显偏慢的 straggler。 单机多卡时 rank 列来自 torch.distributed；多机时使用 global.* 并带 _host/_rank 标签。
Parameters
step_window(integer, default20): Include collectives from the last N global_stepsuse_global(boolean, defaultTrue): Query global.python.comm_collective for cross-node fan-out
Related skills
- 某 rank 持续最慢 → 检查该节点 GPU/网络/数据: skill: gpu_pressure
- 栈卡在 collective → skill: training_hang
- 模块级热点 → skill: module_bottleneck (在慢 rank 上 inject)
- 有 NCCL profiler → skill: nccl_culprit_victim
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 32 lines · 13 tokens per session scan A eb9e438b1a5f
slow_rank is a skill published in the GitHub repository DeepLink-org/probing (11 stars, last pushed 2d ago), licensed Apache-2.0. It adds 13 tokens to every session and 329 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
Market Research Brief
Generate a structured market research brief from a feature idea and supplied business context: market landscape, competitor intelligence, opportunity assessment, and a go-to-market recommendation. Use for market, research, competitor, and product launch requests.
Release Notes Generator
Generate professional software release notes from a commit log: classify changes, write user-facing summaries, draft a publishable announcement, and assess release readiness. Use for release, changelog, version, and deploy requests.
Python 数据分析
A set of instructions for using Python, a programming language, and pandas, a library for working with tables of data. It covers analysing files, calculating statistics, grouping data and making charts.
python-idioms
A set of guidelines for writing clear, safe, maintainable Python. It covers common language features, type annotations, standard-library choices, function design, exceptions, and resource handling.
qmt-docs
A Chinese-language guide to developing Python strategies for QMT, 迅投极速策略交易系统, a platform for quantitative trading. It covers strategy code, backtesting, live trading, APIs, examples, and moving strategies from JoinQuant to QMT.
performance-profiling
A guide for improving software speed by measuring where time and resources are actually being used before changing code.