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/redai-infra/relax/debug-hangnpx skills add redai-infra/Relax --skill debug-hanggit clone --depth 1 https://github.com/redai-infra/RelaxWrote 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/redai-infra/relax/debug-hang)<a href="https://agentmods.dev/skills/redai-infra/relax/debug-hang"><img src="https://agentmods.dev/badge/skills/redai-infra/relax/debug-hang.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 | $0.00067 | $0.02695 |
| Opus 5 | $0.00034 | $0.01347 |
| Sonnet 5 | $0.00013 | $0.00539 |
| Haiku 4.5 | $0.00007 | $0.00269 |
Grade A, and why
debug-hang 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 183 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Ray 分布式训练 Hang 问题自动排查
排查流程
Phase 1: 集群状态概览
目标: 确认集群健康状态和资源使用情况
ray status --address <address>
ray job list --address="<address>" | grep RUNNING
关注指标: 节点存活、CPU/GPU 使用率（异常低 → hang）、Pending resource demands、Object store 内存。
Phase 2: 定位阻塞 Tasks
ray list tasks --address="<address>" --filter "JOB_ID=<job_id>" --filter "state=RUNNING" --format yaml
关键字段: name（业务逻辑）、actor_id、worker_pid（调用栈用）、node_id（py-spy 必须在正确节点执行）。
⚠️ 返回
No resource in the cluster/ 空列表是常态，不是错误。Actor 内部await/time.sleep/dist.barrier等阻塞不会显示为 RUNNING task — actor 主线程一直停在worker.main_loop，业务逻辑跑在后台线程或 asyncio coroutine 里。不要继续试state=SUBMITTED_TO_WORKER/PENDING_NODE_ASSIGNMENT等其他 state，直接跳到 Phase 4 拉 actor 列表 + Phase 3 对 actor PID 跑 py-spy。
Phase 3: 收集调用栈
重要:
py-spy dump --pid <pid>必须在目标进程所在的节点上执行。
# 列出所有节点
ray job submit --working-dir "./" --address="<address>" -- \
python scripts/tools/run_on_each_ray_node.py --list
# 在指定节点执行 py-spy（推荐）
ray job submit --working-dir "./" --address="<address>" -- \
python scripts/tools/run_on_each_ray_node.py -n <node_id_or_ip> "py-spy dump --pid <pid>"
# 在所有 GPU 节点执行（单节点集群适用）
ray job submit --working-dir "./" --address="<address>" -- \
python scripts/tools/run_on_each_ray_node.py "py-spy dump --pid <pid>"
重点关注: 主线程阻塞点、后台线程状态、[Has the GIL] 标记。
⚠️ 反模式（实际踩过的坑）
| 反模式 | 现象 | 正确做法 |
|---|---|---|
本地 py-spy dump --pid <remote_pid> |
Error: No such file or directory (os error 2) |
PID 来自远端 actor 的 worker_pid，必须通过 ray job submit + run_on_each_ray_node.py -n <node_id> 在对应节点执行 |
RAY_ADDRESS=... python scripts/tools/run_on_each_ray_node.py --list |
启了新的本地 Ray 实例，看不到目标集群 | run_on_each_ray_node.py 内部 ray.init() 不带 address，env var 不生效；必须 ray job submit --address="<addr>" -- python scripts/tools/run_on_each_ray_node.py --list |
ray job submit -- bash -c 'for pid in 1 2 3; do py-spy --pid $pid; done' |
py-spy 收到空的 --pid，$pid 被 ray 的引号嵌套吃掉 |
把循环写到 pyspy_dump.sh 文件，再 ray job submit --working-dir "./" -- bash pyspy_dump.sh（脚本随 working-dir 一起上传） |
一个 PID 一个 ray job submit |
每次 ~30-60s 启动开销 × N 个 PID | 写一个脚本文件循环 dump 所有 PID，单次 ray job submit 跑完 |
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.
- 5d ago First seen · 183 lines · 67 tokens per session scan A b0d50bbdaf72
debug-hang is a skill published in the GitHub repository redai-infra/Relax (580 stars, last pushed 7d ago), licensed Apache-2.0. It adds 67 tokens to every session and 2,695 once invoked, about $0.0003 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
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
openrlhf-training
High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
openrlhf-training
High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
openrlhf-training
High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.