Loading...
Loading...
Found 21 Skills
Rigor Debug / Rigor Audit skill for deep learning research work. Use when the user pastes a traceback, terminal error, CUDA OOM, checkpoint load failure, shape mismatch, NaN loss symptom, or training failure and wants conservative diagnosis before any patching, with debug fixes clearly separated from research contributions. Do not use for broad refactoring, speculative adaptation, automatic exploratory patching, or general repository familiarization.
Diagnose, classify, triage, and repair failing Momentic tests with MCP run tools, the Momentic CLI, and manual run artifacts. Use when a developer asks what happened on a branch, DevX or on-call asks why main is red, or the user wants to inspect classifications, de-flake quarantined or recovered tests, reduce retries, re-classify runs, run AI triage, or repair failures.
Use this skill when the user wants to debug, diagnose, or systematically iterate on an experiment that already exists, or when they need a structured experiment log for tracking runs, hypotheses, failures, results, and next steps during active research. Apply it to underperforming methods, training that will not converge, regressions after a change, inconsistent results across datasets, aimless experimentation without progress, and questions like 'why doesn't this work?', 'no progress after many attempts', or 'how should I investigate this failure?'. Also use it for setting up practical experiment logging/record-keeping that supports debugging and iteration. Do not use it for designing a brand-new experiment pipeline or full experiment program (use experiment-pipeline), generating research ideas, fixing isolated coding/syntax errors, or writing retrospective summaries into research memory/notes/knowledge bases.
Diagnose pytest or CI failures, identify root cause, and implement the minimal fix. Use when tests fail or CI reports errors.
Evaluates RAG (Retrieval-Augmented Generation) pipeline quality across retrieval and generation stages. Measures precision, recall, MRR for retrieval; groundedness, completeness, and hallucination rate for generation. Diagnoses failure root causes and recommends chunk, retrieval, and prompt improvements. Triggers on: "audit RAG", "RAG quality", "evaluate retrieval", "hallucination detection", "retrieval precision", "why is RAG failing", "RAG diagnosis", "retrieval quality", "RAG evaluation", "chunk quality", "RAG pipeline review", "grounding check". Use this skill when diagnosing or evaluating a RAG pipeline's quality.
(NS) Root-cause diagnosis from errors, logs, stack traces, failing tests, or CI — entry priority 4 when the user wants diagnosis WITHOUT implement ("why does X break?"). Use proactively on unclear test/CI failures if they have not asked to fix yet. Do NOT use for implement/fix (ns-coder), version scope (ns-spec-driven), GitLab ISSUE_URL execution, or pre-merge review (ns-reviewer).
Diagnose and resolve Celigo flow failures -- total failures, partial errors, stuck jobs, empty runs, and performance issues. Use when a flow is failing, producing errors, returning no data, or running slowly.
Fetch CI build results and diagnose failures. Auto-detects provider from project files or URLs. Supports GitHub Actions, Buildkite, and CircleCI.
Detects MSBuild projects with conflicting OutputPath or IntermediateOutputPath. Only activate in MSBuild/.NET build context. USE FOR: builds failing with 'Cannot create a file when that file already exists', 'The process cannot access the file because it is being used by another process', intermittent build failures that succeed on retry, missing outputs in multi-project builds, multi-targeting builds where project.assets.json conflicts. Diagnoses when multiple projects or TFMs write to the same bin/obj directories due to shared OutputPath, missing AppendTargetFrameworkToOutputPath, or extra global properties like PublishReadyToRun creating redundant evaluations. DO NOT USE FOR: file access errors unrelated to MSBuild (OS-level locking), single-project single-TFM builds, non-MSBuild build systems. INVOKES: dotnet msbuild binlog replay, grep for output path analysis.
Provides guidance for troubleshooting Apache Airflow DAGs (failed DAG runs and task instances) in Managed Service for Apache Airflow (MSAA; formerly Cloud Composer). Use when figuring out reasons for DAG run or task instance failures. Don't use when looking for overall recommendations for Managed Airflow environment performance.
How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.
Validate simulations across three stages — run pre-flight checks on configuration files (parameter ranges, required fields, disk space), monitor runtime logs for residual growth, NaN/Inf, and adaptive dt collapse, and perform post-flight validation of results (physical bounds, mass/energy conservation, convergence). Diagnose failed simulations with probable-cause analysis and recommended fixes. Use when preparing to launch a simulation, checking whether a running job is healthy, verifying that finished results are trustworthy, or debugging a crash or blow-up, even if the user only says "my simulation crashed" or "can I trust these results."