acme_health
01Skill Claude CodeCodex
Demo skill from probing-acme vendor package. Use to verify pip-installed extension discovery (skills + magics).
11 3d ago A 26 tokens
original Apache-2.0
11Stars on the repository
15Mods indexed here, across every type
3d agoLast push, which is what freshness is scored on
Apache-2.0Licence, which decides whether bodies are shown
Skill Claude CodeCodex
Rank collective ops by total and p99 latency.
11 3d ago A 15 tokens
original Apache-2.0
crash_triage
03Skill Claude CodeCodex
Crash triage: grep PROBING CRASH in rank logs; read spill JSON under {PROBINGDATADIR}/crash/ /latest.json.
11 3d ago A 38 tokens
original Apache-2.0
gpu_pressure
04Skill Claude CodeCodex
GPU memory and utilization headroom.
11 3d ago A 10 tokens
original Apache-2.0
Skill Claude CodeCodex
One-shot health check: CPU, GPU, tables, torch step progress, TorchProbe overhead, cluster nodes, NCCL profiler health.
11 3d ago A 31 tokens
original Apache-2.0
job_health
06Skill Claude CodeCodex
Job-level health: cluster alive, step lag, slowdown proxy, comm trend.
11 3d ago A 19 tokens
original Apache-2.0
memory_leak
07Skill Claude CodeCodex
Detect monotonic GPU memory growth across training steps.
11 3d ago A 14 tokens
original Apache-2.0
Skill Claude CodeCodex
Find slowest PyTorch modules in recent steps.
11 3d ago A 15 tokens
original Apache-2.0
Skill Claude CodeCodex
NCCL proxy wait decomposition — culprit (sendgpuwait) vs victim (recvwait).
11 3d ago A 27 tokens
original Apache-2.0
Skill Claude CodeCodex
Find ranks that are slow on many steps (worstfraction), not just once.
11 3d ago A 22 tokens
original Apache-2.0
slow_rank
11Skill Claude CodeCodex
Find straggler ranks via collective latency imbalance.
11 3d ago A 13 tokens
original Apache-2.0
sre_triage
12Skill Claude CodeCodex
SRE first-response triage for distributed training incidents. Automates the manual checks from PyTorch/NCCL debugging runbooks.
11 3d ago A 31 tokens
original Apache-2.0
Skill Claude CodeCodex
Training stalled — capture stacks and thread/blocking hints.
11 3d ago A 15 tokens
original Apache-2.0
Skill Claude CodeCodex
Diagnose PyTorch NCCL watchdog timeouts with Flight Recorder collective sequence alignment.
11 3d ago A 20 tokens
original Apache-2.0