[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78SFPDWGR3K6SBTCZAH4M4P":3,"skill-same:01M78SFPDWGR3K6SBTCZAH4M4P":83},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":33,"install":34,"resources":46},"01M78SFPDWGR3K6SBTCZAH4M4P","evaluation-and-logs",null,"Inspect AlpaSim ASL rollouts, extract camera streams, evaluate and aggregate metrics, re-evaluate completed runs, render diagnostic videos, and analyze trajectories, maps, and geometry without producing new simulations.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: evaluation-and-logs\ndescription: \"Inspect AlpaSim ASL rollouts, extract camera streams, evaluate and\n  aggregate metrics, re-evaluate completed runs, render diagnostic videos, and\n  analyze trajectories, maps, and geometry without producing new simulations.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# Evaluation and logs\n\nUse this sub-skill when the task starts with an existing AlpaSim run, `.asl` log,\nrollout directory, metrics parquet, evaluation configuration, trajectory, map,\nor diagnostic video. It is for reading, scoring, aggregating, and explaining\noutputs—not for starting the simulator or fixing controller, physics, traffic,\nrenderer, or deployment services.\n\n## Route first\n\n- **Runtime production, replay service orchestration, daemon health, or timing\n  failures:** route to `runtime-services`.\n- **Wizard setup, scene\u002Fmodel acquisition, Docker\u002FSlurm deployment, or run\n  creation:** route to `simulation-wizard`.\n- **Driver policy\u002Fmodel implementation:** route to `drivers-and-plugins`.\n- **Controller, physics, or traffic implementation:** route to\n  `control-physics-traffic`.\n- **Proto compilation, generated stubs, or contributor tooling:** route to\n  `grpc-and-developer-tools`.\n\nKeep this sub-skill for the evidence and post-run analysis boundary.\n\n## Choose the smallest workflow\n\n1. Identify the run root and whether it is a single job or an array-job parent.\n   A single job has `eval-config.yaml` at its root; an array parent has child\n   job directories containing that file. Confirm `wizard-config.yaml` is\n   present before using automatic re-evaluation.\n2. Locate complete rollouts under `rollouts\u002F\u003Cscene-or-clip-id>\u002F\u003Crollout-id>\u002F`.\n   Treat a rollout as complete only when its `_complete` marker exists. Expect\n   `rollout.asl`, optional `metrics.parquet`, and optional `.mp4` files there.\n3. For a quick log diagnosis, read the [ASL format reference](references\u002Fasl-format.md)\n   and use the bundled `scripts\u002Fprint_asl.py`. Start with `--just-types`, a\n   narrow `--start\u002F--end` range, and selected\n   `--message-types`; payloads containing images or video-model bytes are\n   redacted before printing. Use `--strict` when truncation must fail loudly.\n4. For camera inspection, use `scripts\u002Fasl_to_frames.py` with a quoted recursive\n   glob. Choose `--format frames` for bounded file inspection or `--format mp4`\n   when ffmpeg\u002Fimageio is available. Keep `--max-files` and\n   `--max-concurrency` bounded for large run trees.\n5. For post-evaluation, run the documented module form in\n   [evaluation workflows](references\u002Fevaluation-workflows.md). Read\n   [geometry and utilities](references\u002Fgeometry-and-utilities.md) before\n   comparing trajectories or maps. Evaluation\n   requires the resolved YAML, complete ASLs, and a matching USDZ glob when\n   map-backed metrics or videos are enabled. It writes per-rollout parquet\n   beside each ASL.\n6. Aggregate only after all intended jobs have metrics. The aggregation step\n   discovers `rollouts\u002F**\u002Fmetrics.parquet`, writes the aggregate tree, and may\n   create relative video links. For a prior run, prefer `python -m eval.reeval`\n   or the equivalent installed re-evaluation command; never submit scheduler\n   work from this sub-skill without an explicit user request.\n\n## Read the data correctly\n\n- ASL is a big-endian, four-byte size-delimited protobuf stream of `LogEntry`\n  messages. The first meaningful message should be `rollout_metadata`; actor\n  poses are timestamped and include `EGO`. Requests\u002Freturns, camera images,\n  routes, and traffic predictions are additional evidence.\n- Use timestamps (`timestamp_us`) rather than list positions when correlating\n  poses, plans, camera frames, and metrics. `async_read_pb_log()` is tolerant\n  by default: a short final frame logs a warning and stops. Set\n  `raise_on_malformed=True` for diagnosis or validation.\n- Evaluation loads ASL through `load_scenario_eval_input_from_asl()`, then\n  `ScenarioEvaluator(EvalConfig).evaluate(...)` returns per-timestep\n  `MetricReturn` values, aggregated values, and a Polars metrics dataframe.\n  Missing rollout metadata, actor definitions, ground truth, or required\n  response\u002Fcamera data can make a metric unavailable or the input invalid.\n- Map-dependent offroad and video analysis need a discoverable USDZ artifact\n  and compatible map data. Without a vector map, aggregation supplies an\n  `offroad=0.0` fallback and records a warning; do not interpret that as proof\n  that offroad was measured.\n\n## Interpret outputs\n\nUse [metrics and videos](references\u002Fmetrics-and-videos.md) for the complete\nmetric and output contract. In short: per-rollout metrics live next to the ASL;\n`aggregate\u002Fmetrics_results.txt`, `metrics_results.parquet`,\n`metrics_unprocessed.parquet`, `results-summary.json`, and\n`metrics_results.png` summarize a run. Videos are named from clip, rollout,\ncamera, and layout identifiers and are organized under `aggregate\u002Fvideos\u002F`.\nInspect the raw parquet and the modifiers recorded in the text summary before\ncomparing runs.\n\n## Programmatic analysis\n\n- Use `alpasim_utils.logs.read_trajectory()` for a scene name plus the EGO\n  trajectory extracted from actor-pose messages. It returns `None` when there\n  are no usable actor poses or no metadata scene id.\n- Use `alpasim_utils.geometry` for gRPC conversion, `Pose`, `Polyline`, and\n  timestamped `Trajectory`. Internally, quaternion arrays are `(x, y, z, w)`;\n  gRPC fields are addressed as `(w, x, y, z)`. Preserve the rig-to-AABB-center\n  transform from rollout metadata before comparing map or vehicle geometry.\n- Use `Trajectory.interpolate_pose()` only inside its time range (a one-pose\n  trajectory accepts its exact timestamp); sort or validate timestamps before\n  constructing trajectories. Use Shapely objects for 2-D intersections and\n  distances, and state when z is intentionally ignored.\n- Treat the Rust-backed `utils_rs` types as an optional native boundary: verify\n  that the extension and generated protobufs import before relying on geometry\n  helpers. The Python evaluation path is not a substitute for a missing map,\n  renderer, ffmpeg, or model asset.\n\n## Safe checks and failure recovery\n\nRun `python -m ... --help`, import checks, and bounded fixture checks before a\nlarge evaluation. Do not run the original checkout's setup, asset download,\nDocker build, Slurm submission, or credentialed model acquisition from here.\nUse [troubleshooting](references\u002Ftroubleshooting.md) for install\u002Fimport,\noptional backend, data\u002Fconfig, CLI\u002FAPI, and workflow failures. When evidence\nis incomplete, report the exact missing message, asset, marker, config field,\nor backend rather than filling in a metric or claiming a successful run.\n\n## Bundled helpers\n\n- `scripts\u002Fprint_asl.py`: bounded, redacted protobuf inspection; safe and\n  read-only.\n- `scripts\u002Fasl_to_frames.py`: bounded extraction to images or MP4; writes only\n  under the requested output directory and never downloads data.\n- `scripts\u002Fcheck_run_metadata.py`: read-only single\u002Farray-job layout and ASL\n  metadata checker; it does not load USDZ maps or rewrite configs.\n\nThe package's `print-asl` and `asl-to-frames` console launchers currently\nrequest a `main` symbol that the installed modules do not define. Use\n`python -m alpasim_utils.print_asl` and `python -m alpasim_utils.asl_to_frames`\n(or these bundled helpers) until packaging is corrected. The evaluation,\naggregation, and re-evaluation module entry points expose `main` and their\n`python -m` forms are the portable fallback.\n","# Evaluation and logs\n\nUse this sub-skill when the task starts with an existing AlpaSim run, `.asl` log,\nrollout directory, metrics parquet, evaluation configuration, trajectory, map,\nor diagnostic video. It is for reading, scoring, aggregating, and explaining\noutputs—not for starting the simulator or fixing controller, physics, traffic,\nrenderer, or deployment services.\n\n## Route first\n\n- **Runtime production, replay service orchestration, daemon health, or timing\n  failures:** route to `runtime-services`.\n- **Wizard setup, scene\u002Fmodel acquisition, Docker\u002FSlurm deployment, or run\n  creation:** route to `simulation-wizard`.\n- **Driver policy\u002Fmodel implementation:** route to `drivers-and-plugins`.\n- **Controller, physics, or traffic implementation:** route to\n  `control-physics-traffic`.\n- **Proto compilation, generated stubs, or contributor tooling:** route to\n  `grpc-and-developer-tools`.\n\nKeep this sub-skill for the evidence and post-run analysis boundary.\n\n## Choose the smallest workflow\n\n1. Identify the run root and whether it is a single job or an array-job parent.\n   A single job has `eval-config.yaml` at its root; an array parent has child\n   job directories containing that file. Confirm `wizard-config.yaml` is\n   present before using automatic re-evaluation.\n2. Locate complete rollouts under `rollouts\u002F\u003Cscene-or-clip-id>\u002F\u003Crollout-id>\u002F`.\n   Treat a rollout as complete only when its `_complete` marker exists. Expect\n   `rollout.asl`, optional `metrics.parquet`, and optional `.mp4` files there.\n3. For a quick log diagnosis, read the [ASL format reference](references\u002Fasl-format.md)\n   and use the bundled `scripts\u002Fprint_asl.py`. Start with `--just-types`, a\n   narrow `--start\u002F--end` range, and selected\n   `--message-types`; payloads containing images or video-model bytes are\n   redacted before printing. Use `--strict` when truncation must fail loudly.\n4. For camera inspection, use `scripts\u002Fasl_to_frames.py` with a quoted recursive\n   glob. Choose `--format frames` for bounded file inspection or `--format mp4`\n   when ffmpeg\u002Fimageio is available. Keep `--max-files` and\n   `--max-concurrency` bounded for large run trees.\n5. For post-evaluation, run the documented module form in\n   [evaluation workflows](references\u002Fevaluation-workflows.md). Read\n   [geometry and utilities](references\u002Fgeometry-and-utilities.md) before\n   comparing trajectories or maps. Evaluation\n   requires the resolved YAML, complete ASLs, and a matching USDZ glob when\n   map-backed metrics or videos are enabled. It writes per-rollout parquet\n   beside each ASL.\n6. Aggregate only after all intended jobs have metrics. The aggregation step\n   discovers `rollouts\u002F**\u002Fmetrics.parquet`, writes the aggregate tree, and may\n   create relative video links. For a prior run, prefer `python -m eval.reeval`\n   or the equivalent installed re-evaluation command; never submit scheduler\n   work from this sub-skill without an explicit user request.\n\n## Read the data correctly\n\n- ASL is a big-endian, four-byte size-delimited protobuf stream of `LogEntry`\n  messages. The first meaningful message should be `rollout_metadata`; actor\n  poses are timestamped and include `EGO`. Requests\u002Freturns, camera images,\n  routes, and traffic predictions are additional evidence.\n- Use timestamps (`timestamp_us`) rather than list positions when correlating\n  poses, plans, camera frames, and metrics. `async_read_pb_log()` is tolerant\n  by default: a short final frame logs a warning and stops. Set\n  `raise_on_malformed=True` for diagnosis or validation.\n- Evaluation loads ASL through `load_scenario_eval_input_from_asl()`, then\n  `ScenarioEvaluator(EvalConfig).evaluate(...)` returns per-timestep\n  `MetricReturn` values, aggregated values, and a Polars metrics dataframe.\n  Missing rollout metadata, actor definitions, ground truth, or required\n  response\u002Fcamera data can make a metric unavailable or the input invalid.\n- Map-dependent offroad and video analysis need a discoverable USDZ artifact\n  and compatible map data. Without a vector map, aggregation supplies an\n  `offroad=0.0` fallback and records a warning; do not interpret that as proof\n  that offroad was measured.\n\n## Interpret outputs\n\nUse [metrics and videos](references\u002Fmetrics-and-videos.md) for the complete\nmetric and output contract. In short: per-rollout metrics live next to the ASL;\n`aggregate\u002Fmetrics_results.txt`, `metrics_results.parquet`,\n`metrics_unprocessed.parquet`, `results-summary.json`, and\n`metrics_results.png` summarize a run. Videos are named from clip, rollout,\ncamera, and layout identifiers and are organized under `aggregate\u002Fvideos\u002F`.\nInspect the raw parquet and the modifiers recorded in the text summary before\ncomparing runs.\n\n## Programmatic analysis\n\n- Use `alpasim_utils.logs.read_trajectory()` for a scene name plus the EGO\n  trajectory extracted from actor-pose messages. It returns `None` when there\n  are no usable actor poses or no metadata scene id.\n- Use `alpasim_utils.geometry` for gRPC conversion, `Pose`, `Polyline`, and\n  timestamped `Trajectory`. Internally, quaternion arrays are `(x, y, z, w)`;\n  gRPC fields are addressed as `(w, x, y, z)`. Preserve the rig-to-AABB-center\n  transform from rollout metadata before comparing map or vehicle geometry.\n- Use `Trajectory.interpolate_pose()` only inside its time range (a one-pose\n  trajectory accepts its exact timestamp); sort or validate timestamps before\n  constructing trajectories. Use Shapely objects for 2-D intersections and\n  distances, and state when z is intentionally ignored.\n- Treat the Rust-backed `utils_rs` types as an optional native boundary: verify\n  that the extension and generated protobufs import before relying on geometry\n  helpers. The Python evaluation path is not a substitute for a missing map,\n  renderer, ffmpeg, or model asset.\n\n## Safe checks and failure recovery\n\nRun `python -m ... --help`, import checks, and bounded fixture checks before a\nlarge evaluation. Do not run the original checkout's setup, asset download,\nDocker build, Slurm submission, or credentialed model acquisition from here.\nUse [troubleshooting](references\u002Ftroubleshooting.md) for install\u002Fimport,\noptional backend, data\u002Fconfig, CLI\u002FAPI, and workflow failures. When evidence\nis incomplete, report the exact missing message, asset, marker, config field,\nor backend rather than filling in a metric or claiming a successful run.\n\n## Bundled helpers\n\n- `scripts\u002Fprint_asl.py`: bounded, redacted protobuf inspection; safe and\n  read-only.\n- `scripts\u002Fasl_to_frames.py`: bounded extraction to images or MP4; writes only\n  under the requested output directory and never downloads data.\n- `scripts\u002Fcheck_run_metadata.py`: read-only single\u002Farray-job layout and ASL\n  metadata checker; it does not load USDZ maps or rewrite configs.\n\nThe package's `print-asl` and `asl-to-frames` console launchers currently\nrequest a `main` symbol that the installed modules do not define. Use\n`python -m alpasim_utils.print_asl` and `python -m alpasim_utils.asl_to_frames`\n(or these bundled helpers) until packaging is corrected. The evaluation,\naggregation, and re-evaluation module entry points expose `main` and their\n`python -m` forms are the portable fallback.\n",{"name":5,"license":23,"metadata":24,"description":7,"disable-model-invocation":26},"Apache 2.0",{"disco-role":25},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":26},[36,39,42],{"tool":37,"command":38},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-logs",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-logs",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-logs","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002FSKILL.md",[47,53,57,61,65,69,75,79],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fasl-format.md","REFERENCE","参考文档",3899,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Freferences\u002Fasl-format.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Fevaluation-workflows.md",5091,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Freferences\u002Fevaluation-workflows.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Fgeometry-and-utilities.md",4130,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Freferences\u002Fgeometry-and-utilities.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fmetrics-and-videos.md",5364,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Freferences\u002Fmetrics-and-videos.md",{"path":66,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Ftroubleshooting.md",6492,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Freferences\u002Ftroubleshooting.md",{"path":70,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":73,"sourceHtmlUrl":74},"scripts\u002Fasl_to_frames.py","SCRIPT","脚本",8860,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Fscripts\u002Fasl_to_frames.py",{"path":76,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":77,"sourceHtmlUrl":78},"scripts\u002Fcheck_run_metadata.py",5175,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Fscripts\u002Fcheck_run_metadata.py",{"path":80,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":81,"sourceHtmlUrl":82},"scripts\u002Fprint_asl.py",3587,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falpasim\u002Fsub-skills\u002Fevaluation-and-logs\u002Fscripts\u002Fprint_asl.py",[84,91,98,105,112,119],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":106,"name":107,"nameZh":6,"description":108,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":109,"categories":110,"keywords":111,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":118,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":120,"name":121,"nameZh":6,"description":122,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":123,"categories":124,"keywords":125,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]