[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"extension-skill-lllllllama-run-train-en":3,"guides-for-lllllllama-run-train":500,"similar-k17bmxf37ewg3r45z7ef99p7z986mf5w-en":501},{"_creationTime":4,"_id":5,"children":6,"community":7,"display":9,"evaluation":15,"identity":243,"isFallback":228,"parentExtension":248,"providers":249,"relations":254,"repo":256,"tags":496,"workflow":497},1778692396032.7788,"k17bmxf37ewg3r45z7ef99p7z986mf5w",[],{"reviewCount":8},0,{"description":10,"installMethods":11,"name":13,"sourceUrl":14},"Trusted-lane training execution skill for deep learning research repositories. Use when a documented or selected training command should be run conservatively for startup verification, short-run verification, full kickoff, or resume, with status, checkpoint, and metric capture written to standardized `train_outputs/`. Do not use for environment setup, exploratory sweeps, speculative idea implementation, or end-to-end orchestration.",{"claudeCode":12},"lllllllama/ai-paper-reproduction-skill","run-train","https://github.com/lllllllama/ai-paper-reproduction-skill",{"_creationTime":16,"_id":17,"extensionId":5,"locale":18,"result":19,"trustSignals":226,"workflow":241},1778692620717.4077,"kn7bgrtjqt8r9kdqe4xctfk7hs86m502","en",{"checks":20,"evaluatedAt":193,"extensionSummary":194,"features":195,"nonGoals":201,"promptVersionExtension":206,"promptVersionScoring":207,"purpose":208,"rationale":209,"score":210,"summary":211,"tags":212,"targetMarket":219,"tier":220,"useCases":221},[21,26,29,32,36,39,44,48,51,54,58,62,65,69,72,75,78,81,84,87,91,95,99,104,108,111,115,118,122,125,128,131,134,137,140,144,148,151,154,158,161,164,167,170,174,177,180,183,186,190],{"category":22,"check":23,"severity":24,"summary":25},"Practical Utility","Problem relevance","pass","The description clearly names a concrete problem: running deep learning training commands conservatively for verification and status capture in research repositories.",{"category":22,"check":27,"severity":24,"summary":28},"Unique selling proposition","The skill offers a specialized, conservative execution lane for training commands with structured output capture, which goes beyond default LLM behavior.",{"category":22,"check":30,"severity":24,"summary":31},"Production readiness","The skill appears ready for production use, handling training execution, status capture, and output writing within the specified domain.",{"category":33,"check":34,"severity":24,"summary":35},"Scope","Single responsibility principle","The skill focuses solely on executing and monitoring training commands, adhering to a single responsibility.",{"category":33,"check":37,"severity":24,"summary":38},"Description quality","The description accurately reflects the skill's purpose of conservative training execution and output capture, including clear use cases and boundaries.",{"category":40,"check":41,"severity":42,"summary":43},"Invocation","Scoped tools","not_applicable","This is a skill that executes a single command, not a collection of tools with individual scopes.",{"category":45,"check":46,"severity":24,"summary":47},"Documentation","Configuration & parameter reference","Key parameters like --repo, --command, --timeout, --run-mode, and others are documented in the script's help text.",{"category":33,"check":49,"severity":42,"summary":50},"Tool naming","As a skill, there are no user-facing tools/commands to evaluate for naming conventions.",{"category":33,"check":52,"severity":24,"summary":53},"Minimal I/O surface","The skill takes specific command-line arguments and outputs structured JSON, adhering to a minimal I/O surface.",{"category":55,"check":56,"severity":24,"summary":57},"License","License usability","The extension is licensed under the MIT license, which is a permissive open-source license.",{"category":59,"check":60,"severity":24,"summary":61},"Maintenance","Commit recency","The last commit was on May 9, 2026, which is within the last 3 months.",{"category":59,"check":63,"severity":42,"summary":64},"Dependency Management","The script appears to use only standard Python libraries and does not list any external third-party dependencies requiring management.",{"category":66,"check":67,"severity":24,"summary":68},"Security","Secret Management","The script does not appear to handle or expose secrets; it focuses on executing a command and capturing output.",{"category":66,"check":70,"severity":24,"summary":71},"Injection","The script uses `shlex.split` to parse the command, mitigating shell injection risks for the command string itself. It doesn't fetch external data.",{"category":66,"check":73,"severity":24,"summary":74},"Transitive Supply-Chain Grenades","The script does not fetch external code or data at runtime; all dependencies are local Python libraries.",{"category":66,"check":76,"severity":24,"summary":77},"Sandbox Isolation","The script operates within the provided repository path and uses standard subprocess execution, respecting sandbox boundaries.",{"category":66,"check":79,"severity":24,"summary":80},"Sandbox escape primitives","No detached process spawns or deny-retry loops were detected in the script.",{"category":66,"check":82,"severity":24,"summary":83},"Data Exfiltration","The script does not perform any outbound calls to submit data and only outputs structured JSON locally.",{"category":66,"check":85,"severity":24,"summary":86},"Hidden Text Tricks","The script code and its associated documentation do not contain any hidden text tricks or obfuscation.",{"category":88,"check":89,"severity":24,"summary":90},"Hooks","Opaque code execution","The Python script is plain, readable source code and does not involve obfuscation techniques.",{"category":92,"check":93,"severity":24,"summary":94},"Portability","Structural Assumption","The script correctly uses relative paths and assumes a standard repository structure, referencing the provided `--repo` argument.",{"category":96,"check":97,"severity":24,"summary":98},"Trust","Issues Attention","There are 0 issues opened and 0 closed in the last 90 days, indicating no recent activity but also no pending issues.",{"category":100,"check":101,"severity":102,"summary":103},"Versioning","Release Management","warning","The script itself does not have a version number, and the repository install instructions primarily reference installing from main (`npx skills add ... --all` or `... --skill ...`), making it difficult to pin a specific version of this script.",{"category":105,"check":106,"severity":24,"summary":107},"Execution","Validation","The script uses `shlex.split` for command parsing and `argparse` for argument validation, ensuring basic input integrity.",{"category":66,"check":109,"severity":24,"summary":110},"Unguarded Destructive Operations","The script's primary function is executing a user-provided command, but it does not perform destructive operations itself without user invocation.",{"category":112,"check":113,"severity":24,"summary":114},"Code Execution","Error Handling","The script handles file not found errors, timeouts, and non-zero exit codes gracefully, reporting them in the output JSON.",{"category":112,"check":116,"severity":24,"summary":117},"Logging","The script outputs a structured JSON payload containing execution details and logs to stdout, serving as an audit record.",{"category":119,"check":120,"severity":42,"summary":121},"Compliance","GDPR","The skill only executes commands and captures output; it does not process personal data.",{"category":119,"check":123,"severity":24,"summary":124},"Target market","The skill is a general-purpose training execution utility and has no regional or jurisdictional logic, making it global.",{"category":92,"check":126,"severity":24,"summary":127},"Runtime stability","The script uses standard Python libraries and practices, ensuring cross-platform compatibility on systems with Python 3.",{"category":45,"check":129,"severity":24,"summary":130},"README","The README file for the repository provides comprehensive details about the skills, including installation and purpose, and the specific skill's SKILL.md file is clear.",{"category":33,"check":132,"severity":42,"summary":133},"Tool surface size","This is a single skill that executes a command, not an extension exposing multiple tools.",{"category":40,"check":135,"severity":42,"summary":136},"Overlapping near-synonym tools","As a single skill, there are no overlapping tools to evaluate.",{"category":45,"check":138,"severity":24,"summary":139},"Phantom features","All advertised features in the SKILL.md and README are implemented in the provided Python script.",{"category":141,"check":142,"severity":24,"summary":143},"Install","Installation instruction","The README provides clear installation instructions using `npx` and offers advanced local install commands.",{"category":145,"check":146,"severity":24,"summary":147},"Errors","Actionable error messages","Errors like 'Executable not found' or 'timeout' are reported with context and a clear indication of failure in the output JSON.",{"category":105,"check":149,"severity":24,"summary":150},"Pinned dependencies","The script relies on standard Python libraries, and the repository includes `scripts/install_skills.py` which suggests dependency management, though a lockfile for the script itself isn't explicit.",{"category":33,"check":152,"severity":42,"summary":153},"Dry-run preview","The skill executes a user-provided command; a dry-run capability would need to be implemented within that command itself, not by this wrapper.",{"category":155,"check":156,"severity":24,"summary":157},"Protocol","Idempotent retry & timeouts","The script enforces a timeout and reports it as a structured error, and the execution itself is designed to be retried if needed by a higher-level orchestrator.",{"category":119,"check":159,"severity":24,"summary":160},"Telemetry opt-in","The script does not emit any telemetry; all output is local JSON.",{"category":40,"check":162,"severity":24,"summary":163},"Precise Purpose","The skill's purpose is precisely defined: conservative execution of selected training commands with structured output capture, explicitly stating when to use and when not to use it.",{"category":40,"check":165,"severity":24,"summary":166},"Concise Frontmatter","The SKILL.md frontmatter is concise, clearly stating the skill's purpose and domain without excessive keywords.",{"category":45,"check":168,"severity":24,"summary":169},"Concise Body","The SKILL.md is concise and delegates deeper material to separate reference files like `training-policy.md` and scripts.",{"category":171,"check":172,"severity":24,"summary":173},"Context","Progressive Disclosure","The SKILL.md refers to external files like `references/training-policy.md` and scripts, demonstrating progressive disclosure.",{"category":171,"check":175,"severity":42,"summary":176},"Forked exploration","This skill is not designed for deep exploration; it executes a specific command and is not expected to set `context: fork`.",{"category":22,"check":178,"severity":24,"summary":179},"Usage examples","The README provides numerous examples for various skills in the repository, including conceptual examples that imply how this skill would be invoked in context.",{"category":22,"check":181,"severity":24,"summary":182},"Edge cases","The script handles common edge cases like command not found, timeouts, and non-zero exit codes, documenting the symptom and recovery path (reported in output).",{"category":112,"check":184,"severity":42,"summary":185},"Tool Fallback","This skill does not rely on external MCP servers or other tools that would require a fallback mechanism.",{"category":187,"check":188,"severity":24,"summary":189},"Safety","Halt on unexpected state","The script is designed to halt execution and report errors (e.g., command not found, timeout, non-zero exit) rather than proceeding in an unexpected state.",{"category":92,"check":191,"severity":24,"summary":192},"Cross-skill coupling","This skill is self-contained and does not implicitly rely on other skills. Its purpose is distinct and well-defined.",1778692620395,"This skill executes a specified training command within a given repository, captures its output and status, and outputs this information in a structured JSON format. It handles startup verification, short runs, full kickoffs, and resumes, while also managing timeouts and errors.",[196,197,198,199,200],"Conservative training command execution","Structured status, checkpoint, and metric capture","Handles startup verification, short-run verification, full kickoff, and resume","Outputs evidence to `train_outputs/`","Error and timeout handling",[202,203,204,205],"Environment setup or asset downloading","Exploratory sweeps or speculative idea implementation","End-to-end orchestration of research goals","Choosing training commands autonomously","3.0.0","4.4.0","To provide a trusted and auditable way to run deep learning training commands conservatively, ensuring verification and structured capture of results.","The skill is well-documented, production-ready, and adheres to security best practices. The only minor finding is the lack of explicit versioning for the script itself, which is common for individual utility scripts within a larger repository.",99,"A robust and well-documented skill for conservative training execution and evidence capture in deep learning research.",[213,214,215,216,217,218],"deep-learning","training","research","verification","monitoring","python","global","community",[222,223,224,225],"Verifying training command startup in a research repository","Running short-duration training for verification purposes","Initiating or resuming full training runs with monitored evidence","Capturing structured logs and checkpoints from training processes",{"codeQuality":227,"collectedAt":229,"documentation":230,"maintenance":233,"security":238,"testCoverage":240},{"hasLockfile":228},false,1778692605425,{"descriptionLength":231,"readmeSize":232},435,22701,{"closedIssues90d":8,"forks":234,"hasChangelog":235,"openIssues90d":8,"pushedAt":236,"stars":237},4,true,1778347974000,75,{"hasNpmPackage":228,"license":239,"smitheryVerified":228},"MIT",{"hasCi":235,"hasTests":235},{"updatedAt":242},1778692620717,{"basePath":244,"githubOwner":245,"githubRepo":246,"locale":18,"slug":13,"type":247},"skills/run-train","lllllllama","ai-paper-reproduction-skill","skill",null,{"evaluate":250,"extract":252},{"promptVersionExtension":206,"promptVersionScoring":207,"score":210,"tags":251,"targetMarket":219,"tier":220},[213,214,215,216,217,218],{"commitSha":253},"HEAD",{"repoId":255},"kd7629v5mqesxwwe9w7qtfgp7d86n6re",{"_creationTime":257,"_id":255,"identity":258,"providers":259,"workflow":492},1778692391648.3123,{"githubOwner":245,"githubRepo":246,"sourceUrl":14},{"classify":260,"discover":486,"github":489},{"commitSha":253,"extensions":261},[262,339,369,381,401,414,427,440,450,464,474],{"basePath":263,"description":264,"displayName":265,"installMethods":266,"rationale":267,"selectedPaths":268,"source":338,"sourceLanguage":18,"type":247},"skills/ai-research-explore","Explore-lane end-to-end orchestrator for the third research scenario: the researcher has already chosen the task family, dataset, benchmark, evaluation method, and provided SOTA references, and wants candidate-only exploration on top of `current_research` with auditable repo understanding, idea gating, and governed experiments written to `explore_outputs/`. Do not use for README-first trusted reproduction, open-ended direction finding, narrow code-only or run-only exploration, passive repo analysis, or implicit experimentation.","ai-research-explore",{"claudeCode":12},"SKILL.md frontmatter at skills/ai-research-explore/SKILL.md",[269,272,275,277,279,281,283,285,288,290,292,294,296,298,300,302,304,306,308,310,312,314,316,318,320,322,324,326,328,330,332,334,336],{"path":270,"priority":271},"SKILL.md","mandatory",{"path":273,"priority":274},"references/ai-research-explore-policy.md","medium",{"path":276,"priority":274},"references/idea-evaluation-framework.md",{"path":278,"priority":274},"references/research-campaign-spec.md",{"path":280,"priority":274},"references/smoke-validation-policy.md",{"path":282,"priority":274},"references/source-mapping-policy.md",{"path":284,"priority":274},"references/sources-naming-policy.md",{"path":286,"priority":287},"scripts/lookup/__init__.py","low",{"path":289,"priority":287},"scripts/lookup/cache_store.py",{"path":291,"priority":287},"scripts/lookup/inventory_writer.py",{"path":293,"priority":287},"scripts/lookup/normalizers.py",{"path":295,"priority":287},"scripts/lookup/providers/__init__.py",{"path":297,"priority":287},"scripts/lookup/providers/arxiv_provider.py",{"path":299,"priority":287},"scripts/lookup/providers/base.py",{"path":301,"priority":287},"scripts/lookup/providers/doi_provider.py",{"path":303,"priority":287},"scripts/lookup/providers/github_provider.py",{"path":305,"priority":287},"scripts/lookup/providers/optional_provider.py",{"path":307,"priority":287},"scripts/lookup/providers/url_provider.py",{"path":309,"priority":287},"scripts/lookup/record_schema.py",{"path":311,"priority":287},"scripts/lookup/repo_extractors.py",{"path":313,"priority":287},"scripts/lookup/source_support.py",{"path":315,"priority":287},"scripts/orchestrate_explore.py",{"path":317,"priority":287},"scripts/passes/__init__.py",{"path":319,"priority":287},"scripts/passes/atomic_idea_decomposition.py",{"path":321,"priority":287},"scripts/passes/candidate_idea_generation.py",{"path":323,"priority":287},"scripts/passes/execution_feasibility.py",{"path":325,"priority":287},"scripts/passes/idea_cards.py",{"path":327,"priority":287},"scripts/passes/idea_ranking.py",{"path":329,"priority":287},"scripts/passes/implementation_fidelity.py",{"path":331,"priority":287},"scripts/passes/improvement_bank.py",{"path":333,"priority":287},"scripts/passes/lookup_sources.py",{"path":335,"priority":287},"scripts/passes/source_mapping.py",{"path":337,"priority":287},"scripts/write_outputs.py","rule",{"basePath":340,"description":341,"displayName":342,"installMethods":343,"rationale":344,"selectedPaths":345,"source":338,"sourceLanguage":18,"type":247},"skills/ai-research-reproduction","Main orchestrator for README-first AI repo reproduction. Use when the user wants an end-to-end, minimal-trustworthy reproduction flow that reads the repository first, selects the smallest documented inference or evaluation target, coordinates intake, setup, trusted execution, optional trusted training, optional repository analysis, and optional paper-gap resolution, enforces conservative patch rules, records evidence assumptions deviations and human decision points, and writes the standardized `repro_outputs/` bundle. Do not use for paper summary, generic environment setup, isolated repo scanning, standalone command execution, silent protocol changes, or broad research assistance outside repository-grounded reproduction.","ai-research-reproduction",{"claudeCode":12},"SKILL.md frontmatter at skills/ai-research-reproduction/SKILL.md",[346,347,349,351,353,355,357,359,361,363,365,367],{"path":270,"priority":271},{"path":348,"priority":287},"assets/COMMANDS.template.md",{"path":350,"priority":287},"assets/LOG.template.md",{"path":352,"priority":287},"assets/PATCHES.template.md",{"path":354,"priority":287},"assets/SUMMARY.template.md",{"path":356,"priority":287},"assets/status.template.json",{"path":358,"priority":274},"references/architecture.md",{"path":360,"priority":274},"references/language-policy.md",{"path":362,"priority":274},"references/output-spec.md",{"path":364,"priority":274},"references/patch-policy.md",{"path":366,"priority":274},"references/research-safety-principles.md",{"path":368,"priority":287},"scripts/orchestrate_repro.py",{"basePath":370,"description":371,"displayName":372,"installMethods":373,"rationale":374,"selectedPaths":375,"source":338,"sourceLanguage":18,"type":247},"skills/analyze-project","Trusted-lane analysis skill for deep learning research repositories. Use when the user wants to read and understand a repository, inspect model structure and training or inference entrypoints, review configs and insertion points, or flag suspicious implementation patterns without modifying code or running heavy jobs. Do not use for active command execution, broad refactoring, speculative code adaptation, or automatic bug fixing.","analyze-project",{"claudeCode":12},"SKILL.md frontmatter at skills/analyze-project/SKILL.md",[376,377,379],{"path":270,"priority":271},{"path":378,"priority":274},"references/analysis-policy.md",{"path":380,"priority":287},"scripts/analyze_project.py",{"basePath":382,"description":383,"displayName":384,"installMethods":385,"rationale":386,"selectedPaths":387,"source":338,"sourceLanguage":18,"type":247},"skills/env-and-assets-bootstrap","Environment and assets sub-skill for README-first AI repo reproduction. Use when the task is specifically to prepare a conservative conda-first environment, checkpoint and dataset path assumptions, cache location hints, and setup notes before any run on a README-documented repository. Do not use for repo scanning, full orchestration, paper interpretation, final run reporting, or generic environment setup that is not tied to a specific reproduction target.","env-and-assets-bootstrap",{"claudeCode":12},"SKILL.md frontmatter at skills/env-and-assets-bootstrap/SKILL.md",[388,389,391,393,395,397,399],{"path":270,"priority":271},{"path":390,"priority":274},"references/assets-policy.md",{"path":392,"priority":274},"references/env-policy.md",{"path":394,"priority":287},"scripts/bootstrap_env.py",{"path":396,"priority":287},"scripts/bootstrap_env.sh",{"path":398,"priority":287},"scripts/plan_setup.py",{"path":400,"priority":287},"scripts/prepare_assets.py",{"basePath":402,"description":403,"displayName":404,"installMethods":405,"rationale":406,"selectedPaths":407,"source":338,"sourceLanguage":18,"type":247},"skills/explore-code","Explore-lane code adaptation skill for deep learning research repositories. Use when the researcher explicitly authorizes exploratory work on an isolated branch or worktree to transplant modules, adapt a backbone, add LoRA or adapter layers, replace a head, or stitch together low-risk migration ideas with summary-only records in `explore_outputs/`. Do not use for end-to-end exploration orchestration on top of `current_research`, trusted baseline reproduction, conservative debugging, environment setup, or default repository analysis.","explore-code",{"claudeCode":12},"SKILL.md frontmatter at skills/explore-code/SKILL.md",[408,409,411,413],{"path":270,"priority":271},{"path":410,"priority":274},"references/explore-policy.md",{"path":412,"priority":287},"scripts/plan_code_changes.py",{"path":337,"priority":287},{"basePath":415,"description":416,"displayName":417,"installMethods":418,"rationale":419,"selectedPaths":420,"source":338,"sourceLanguage":18,"type":247},"skills/explore-run","Explore-lane experimental execution skill for deep learning research repositories. Use when the researcher explicitly authorizes exploratory runs such as small-subset validation, short-cycle guess-and-check, batch sweeps, idle-GPU search, or quick transfer-learning trials, with results summarized in `explore_outputs/`. Do not use for end-to-end exploration orchestration on top of `current_research`, trusted baseline execution, conservative training verification, default routing, or implicit experimentation.","explore-run",{"claudeCode":12},"SKILL.md frontmatter at skills/explore-run/SKILL.md",[421,422,424,426],{"path":270,"priority":271},{"path":423,"priority":274},"references/execution-policy.md",{"path":425,"priority":287},"scripts/plan_variants.py",{"path":337,"priority":287},{"basePath":428,"description":429,"displayName":430,"installMethods":431,"rationale":432,"selectedPaths":433,"source":338,"sourceLanguage":18,"type":247},"skills/minimal-run-and-audit","Trusted-lane execution and reporting skill for README-first AI repo reproduction. Use when the task is specifically to capture or normalize evidence from the selected smoke test or documented inference or evaluation command and write standardized `repro_outputs/` files, including patch notes when repository files changed. Do not use for training execution, initial repo intake, generic environment setup, paper lookup, target selection, or end-to-end orchestration by itself.","minimal-run-and-audit",{"claudeCode":12},"SKILL.md frontmatter at skills/minimal-run-and-audit/SKILL.md",[434,435,437,439],{"path":270,"priority":271},{"path":436,"priority":274},"references/reporting-policy.md",{"path":438,"priority":287},"scripts/run_command.py",{"path":337,"priority":287},{"basePath":441,"description":442,"displayName":443,"installMethods":444,"rationale":445,"selectedPaths":446,"source":338,"sourceLanguage":18,"type":247},"skills/paper-context-resolver","Optional narrow helper skill for README-first AI repo reproduction. Use only when the README and repository files leave a narrow reproduction-critical gap and the task is to resolve a specific paper detail such as dataset split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumption from primary paper sources while recording conflicts. Do not use for general paper summary, repo scanning, environment setup, command execution, title-only paper lookup, or replacing README guidance by default.","paper-context-resolver",{"claudeCode":12},"SKILL.md frontmatter at skills/paper-context-resolver/SKILL.md",[447,448],{"path":270,"priority":271},{"path":449,"priority":274},"references/paper-assisted-reproduction.md",{"basePath":451,"description":452,"displayName":453,"installMethods":454,"rationale":455,"selectedPaths":456,"source":338,"sourceLanguage":18,"type":247},"skills/repo-intake-and-plan","Narrow helper skill for README-first AI repo reproduction. Use when the task is specifically to scan a repository, read the README and common project files, extract documented commands, classify inference, evaluation, and training candidates, and return the smallest trustworthy reproduction plan to the main orchestrator. Do not use for environment setup, asset download, command execution, final reporting, paper lookup, or end-to-end orchestration.","repo-intake-and-plan",{"claudeCode":12},"SKILL.md frontmatter at skills/repo-intake-and-plan/SKILL.md",[457,458,460,462],{"path":270,"priority":271},{"path":459,"priority":274},"references/repo-scan-rules.md",{"path":461,"priority":287},"scripts/extract_commands.py",{"path":463,"priority":287},"scripts/scan_repo.py",{"basePath":244,"description":10,"displayName":13,"installMethods":465,"rationale":466,"selectedPaths":467,"source":338,"sourceLanguage":18,"type":247},{"claudeCode":12},"SKILL.md frontmatter at skills/run-train/SKILL.md",[468,469,471,473],{"path":270,"priority":271},{"path":470,"priority":274},"references/training-policy.md",{"path":472,"priority":287},"scripts/run_training.py",{"path":337,"priority":287},{"basePath":475,"description":476,"displayName":477,"installMethods":478,"rationale":479,"selectedPaths":480,"source":338,"sourceLanguage":18,"type":247},"skills/safe-debug","Trusted-lane debug 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. Do not use for broad refactoring, speculative adaptation, automatic exploratory patching, or general repository familiarization.","safe-debug",{"claudeCode":12},"SKILL.md frontmatter at skills/safe-debug/SKILL.md",[481,482,484],{"path":270,"priority":271},{"path":483,"priority":274},"references/debug-policy.md",{"path":485,"priority":287},"scripts/safe_debug.py",{"sources":487},[488],"manual",{"closedIssues90d":8,"description":490,"forks":234,"license":239,"openIssues90d":8,"pushedAt":236,"readmeSize":232,"stars":237,"topics":491},"",[],{"classifiedAt":493,"discoverAt":494,"extractAt":495,"githubAt":495,"updatedAt":493},1778692395631,1778692391648,1778692393876,[213,217,218,215,214,216],{"evaluatedAt":242,"extractAt":498,"updatedAt":499},1778692396032,1778692750831,[],[502,533,562,589,619,645],{"_creationTime":503,"_id":504,"community":505,"display":506,"identity":512,"providers":516,"relations":526,"tags":529,"workflow":530},1778693180473.1174,"k17fm8t65dw1y28823kj8ce3bn86mgqg",{"reviewCount":8},{"description":507,"installMethods":508,"name":510,"sourceUrl":511},"Azure Monitor Query SDK for Python. Use for querying Log Analytics workspaces and Azure Monitor metrics.\nTriggers: \"azure-monitor-query\", \"LogsQueryClient\", \"MetricsQueryClient\", \"Log Analytics\", \"Kusto queries\", \"Azure metrics\".\n",{"claudeCode":509},"microsoft/agent-skills","azure-monitor-query-py","https://github.com/microsoft/agent-skills",{"basePath":513,"githubOwner":514,"githubRepo":515,"locale":18,"slug":510,"type":247},".github/plugins/azure-sdk-python/skills/azure-monitor-query-py","microsoft","agent-skills",{"evaluate":517,"extract":525},{"promptVersionExtension":206,"promptVersionScoring":207,"score":518,"tags":519,"targetMarket":219,"tier":524},100,[520,217,521,522,523,218],"azure","logs","metrics","sdk","verified",{"commitSha":253},{"parentExtensionId":527,"repoId":528},"k171mfx6atvhq1bkhpky84v4b186n9qd","kd77czgnv00rfjm815pcc5xx5986n5t8",[520,521,522,217,218,523],{"evaluatedAt":531,"extractAt":532,"updatedAt":531},1778695102758,1778693180473,{"_creationTime":534,"_id":535,"community":536,"display":537,"identity":543,"providers":547,"relations":555,"tags":558,"workflow":559},1778695116697.1838,"k17c6fx43mgkj95s4yzww50w5s86nb65",{"reviewCount":8},{"description":538,"installMethods":539,"name":541,"sourceUrl":542},"High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with same code. Use when you want clean training loops with built-in best practices.",{"claudeCode":540},"Orchestra-Research/AI-Research-SKILLs","pytorch-lightning","https://github.com/Orchestra-Research/AI-Research-SKILLs",{"basePath":544,"githubOwner":545,"githubRepo":546,"locale":18,"slug":541,"type":247},"08-distributed-training/pytorch-lightning","Orchestra-Research","AI-Research-SKILLs",{"evaluate":548,"extract":554},{"promptVersionExtension":206,"promptVersionScoring":207,"score":210,"tags":549,"targetMarket":219,"tier":524},[550,551,214,552,553,213],"pytorch","lightning","distributed-training","mlops",{"commitSha":253},{"parentExtensionId":556,"repoId":557},"k17155ws9qc0hw7a568bg79sfd86max8","kd70hj1y80mhra5xm5g188j5n586mg18",[213,552,551,553,550,214],{"evaluatedAt":560,"extractAt":561,"updatedAt":560},1778696329359,1778695116697,{"_creationTime":563,"_id":564,"community":565,"display":566,"identity":572,"providers":576,"relations":583,"tags":585,"workflow":586},1778691799740.4905,"k17c27dcgjsqmxeggb19stv4xn86mf1z",{"reviewCount":8},{"description":567,"installMethods":568,"name":570,"sourceUrl":571},"Deep learning framework (PyTorch Lightning). Organize PyTorch code into LightningModules, configure Trainers for multi-GPU/TPU, implement data pipelines, callbacks, logging (W&B, TensorBoard), distributed training (DDP, FSDP, DeepSpeed), for scalable neural network training.",{"claudeCode":569},"K-Dense-AI/claude-scientific-skills","PyTorch Lightning","https://github.com/K-Dense-AI/claude-scientific-skills",{"basePath":573,"githubOwner":574,"githubRepo":575,"locale":18,"slug":541,"type":247},"scientific-skills/pytorch-lightning","K-Dense-AI","claude-scientific-skills",{"evaluate":577,"extract":581},{"promptVersionExtension":206,"promptVersionScoring":207,"score":518,"tags":578,"targetMarket":219,"tier":524},[550,213,579,218,580],"machine-learning","framework",{"commitSha":253,"license":582},"Apache-2.0",{"repoId":584},"kd79rphh5gexy91xmpxc05h5mh86mm9r",[213,580,579,218,550],{"evaluatedAt":587,"extractAt":588,"updatedAt":587},1778693958717,1778691799740,{"_creationTime":590,"_id":591,"community":592,"display":593,"identity":599,"providers":603,"relations":612,"tags":615,"workflow":616},1778695548458.3782,"k17a4rtftm1z500gdcksks32wx86n9p2",{"reviewCount":8},{"description":594,"installMethods":595,"name":597,"sourceUrl":598},"Design and operate a data integrity monitoring programme based on ALCOA+ principles. Covers detective controls, audit trail review schedules, anomaly detection patterns (off-hours activity, sequential modifications, bulk changes), metrics dashboards, investigation triggers, and escalation matrix definition. Use when establishing a data integrity monitoring programme for GxP systems, preparing for inspections where data integrity is a focus area, after a data integrity incident requiring enhanced monitoring, or when implementing MHRA, WHO, or PIC/S guidance.\n",{"claudeCode":596},"pjt222/agent-almanac","monitor-data-integrity","https://github.com/pjt222/agent-almanac",{"basePath":600,"githubOwner":601,"githubRepo":602,"locale":18,"slug":597,"type":247},"skills/monitor-data-integrity","pjt222","agent-almanac",{"evaluate":604,"extract":611},{"promptVersionExtension":206,"promptVersionScoring":207,"score":518,"tags":605,"targetMarket":219,"tier":524},[606,607,608,609,217,610],"compliance","gxp","data-integrity","alcoa","anomaly-detection",{"commitSha":253},{"parentExtensionId":613,"repoId":614},"k170h0janaa9kwn7cfgfz2ykss86mmh9","kd7aryv63z61j39n2td1aeqkvh86mh12",[609,610,606,608,607,217],{"evaluatedAt":617,"extractAt":618,"updatedAt":617},1778699562914,1778695548458,{"_creationTime":620,"_id":621,"community":622,"display":623,"identity":629,"providers":632,"relations":639,"tags":641,"workflow":642},1778694578248.1062,"k17e56dzsqh7qked458bjbs0e586n21y",{"reviewCount":8},{"description":624,"installMethods":625,"name":627,"sourceUrl":628},"Query Netdata Cloud via its REST API -- metrics, logs (systemd-journal / windows-events / otel-logs), topology graphs (topology:snmp), network flows (flows:netflow), alerts, dynamic configuration (DynCfg), and generic Functions on a node. Use when the user asks about querying Netdata Cloud, fetching metrics from the cloud, querying logs / topology / netflow / sflow / ipfix through Cloud, listing or modifying configurations via DynCfg, calling agent Functions through Cloud, listing spaces/rooms/nodes, or building a curl command against `app.netdata.cloud`. Pairs with the `query-netdata-agents` skill when direct-agent access is needed.",{"claudeCode":626},"netdata/netdata","query-netdata-cloud","https://github.com/netdata/netdata",{"basePath":630,"githubOwner":631,"githubRepo":631,"locale":18,"slug":627,"type":247},"docs/netdata-ai/skills/query-netdata-cloud","netdata",{"evaluate":633,"extract":638},{"promptVersionExtension":206,"promptVersionScoring":207,"score":518,"tags":634,"targetMarket":219,"tier":524},[631,635,217,522,521,636,637],"api","topology","rest",{"commitSha":253},{"repoId":640},"kd70yp91ybn40a638h3hzz6nbd86m2cw",[635,521,522,217,631,637,636],{"evaluatedAt":643,"extractAt":644,"updatedAt":643},1778694825298,1778694578248,{"_creationTime":646,"_id":647,"community":648,"display":649,"identity":655,"providers":659,"relations":667,"tags":669,"workflow":670},1778694240519.7402,"k172jnxq28h65x6zn1p19r731586md2x",{"reviewCount":8},{"description":650,"installMethods":651,"name":653,"sourceUrl":654},"Track skill performance and emerging patterns",{"claudeCode":652},"mshadmanrahman/pm-pilot","meta-observer","https://github.com/mshadmanrahman/pm-pilot",{"basePath":656,"githubOwner":657,"githubRepo":658,"locale":18,"slug":653,"type":247},"skills/productivity/meta-observer","mshadmanrahman","pm-pilot",{"evaluate":660,"extract":666},{"promptVersionExtension":206,"promptVersionScoring":207,"score":518,"tags":661,"targetMarket":219,"tier":524},[217,662,663,664,665],"analytics","productivity","logging","skills",{"commitSha":253},{"repoId":668},"kd728wqst6vwd95ymycxb97nrx86mnsn",[662,664,217,663,665],{"evaluatedAt":671,"extractAt":672,"updatedAt":671},1778694605108,1778694240519]