[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M793C9A0GCKB9QK7M33ZQHMP":3,"skill-same:01M793C9A0GCKB9QK7M33ZQHMP":79},{"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},"01M793C9A0GCKB9QK7M33ZQHMP","robotics-modules",null,"Use PyPose's differentiable robotics modules for discrete dynamics, Bayesian state estimation, LQR\u002FMPC control, IMU preintegration, PnP\u002FICP pose solving, and rotation geodesic loss; choose constructors and tensor contracts from the bundled references, validate shapes\u002Fcovariances\u002Funits, and compose modules without confusing them with primitive manifold math, generic optimization, or geometry metrics.","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\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: robotics-modules\ndescription: \"Use PyPose's differentiable robotics modules for discrete\n  dynamics, Bayesian state estimation, LQR\u002FMPC control, IMU preintegration,\n  PnP\u002FICP pose solving, and rotation geodesic loss; choose constructors and\n  tensor contracts from the bundled references, validate\n  shapes\u002Fcovariances\u002Funits, and compose modules without confusing them with\n  primitive manifold math, generic optimization, or geometry metrics.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# PyPose robotics-modules\n\nUse this skill when a Researcher needs PyPose's high-level robotics modules in\n`pypose.module`: system models (`System`, `LTI`, `LTV`, `NLS`), EKF\u002FUKF\u002FPF,\nLQR\u002FMPC, `IMUPreintegrator`, `EPnP`, `ICP`, or `GeodesicLoss`. The module\nimplementations are differentiable PyTorch `nn.Module`s, but several modules\nalso mutate a clock, integration state, or stopping planner. Treat those state\nand shape contracts as part of the API.\n\n## Scope boundary\n\n- **Included:** constructors, forward signatures, rightmost feature dimensions,\n  leading-batch broadcasting, time and reset behavior, covariance contracts,\n  dynamics\u002Ffilter\u002Fcontrol composition, IMU units and frame conventions, PnP and\n  ICP initialization, and solver-stepper diagnostics.\n- **Excluded:** primitive LieTensor\u002Fmanifold algebra (`lie-tensor`), generic\n  Gauss--Newton\u002FLevenberg--Marquardt optimization (`optimization`), and\n  geometry\u002Fspline\u002Ftrajectory metrics (`geometry-evaluation`). `EPnP(refine=True)`\n  may call its own internal refinement; use the PnP contract here rather than\n  treating that implementation detail as a general optimizer workflow.\n\nRead the smallest relevant reference before writing code:\n\n- [api-reference.md](references\u002Fapi-reference.md) for verified signatures and\n  exact call\u002Freturn contracts.\n- [state-estimation.md](references\u002Fstate-estimation.md) for `System`\u002F`NLS`\n  models and EKF\u002FUKF\u002FPF covariance and shape rules.\n- [control-and-dynamics.md](references\u002Fcontrol-and-dynamics.md) for LTI\u002FLTV,\n  LQR, MPC, horizon layouts, and stepper behavior.\n- [imu-and-geometric-solvers.md](references\u002Fimu-and-geometric-solvers.md) for\n  IMU, EPnP, ICP, and geodesic rotation loss.\n- [troubleshooting.md](references\u002Ftroubleshooting.md) for failure diagnosis and\n  safe recovery.\n\n## Operating workflow\n\n1. **Identify the module boundary.** Decide whether the task needs a dynamic\n   model, a state estimator, a controller, sensor preintegration, a geometric\n   solver, or a rotation loss. Do not replace a module with primitive manifold\n   operations unless the task explicitly needs those primitives.\n2. **Normalize shapes before constructing modules.** Put features in the last\n   dimension, matrices in the last two dimensions, and batch\u002Ftime axes in\n   leading positions. Make `dtype` and `device` consistent across model,\n   covariance, costs, and observations. Do not rely on accidental broadcasting\n   for state versus time axes.\n3. **Build the dependency chain.** A normal filtering chain is\n   `NLS -> (EKF | UKF | PF)`; a control chain is `NLS -> set_refpoint -> LQR`\n   inside `MPC`; a sensor chain is `IMUPreintegrator -> pose\u002Fvelocity\u002Fposition`\n   consumers. `EPnP` and `ICP` each return an `SE3` pose for a separate\n   geometry-registration branch, while `GeodesicLoss` compares rotations.\n4. **Make mutable state explicit.** Reset a dynamic model before replaying a\n   trajectory. Reuse an IMU integrator only when its accumulation semantics are\n   intended. Reuse a `ReduceToBason`-like stepper only after its `reset()`.\n   Avoid concurrent calls through one mutable module.\n5. **Check outputs, not just execution.** Assert expected trailing shapes,\n   finiteness, covariance symmetry\u002Fpositive semidefiniteness within numerical\n   tolerance, plausible units, and an error or cost that improves on a\n   deterministic fixture. Use the bundled scripts for a quick CPU check:\n\n   From the `pypose` skill directory, run:\n\n   ```bash\n   python sub-skills\u002Frobotics-modules\u002Fscripts\u002Ffilter_smoke.py\n   python sub-skills\u002Frobotics-modules\u002Fscripts\u002Ficp_smoke.py\n   ```\n\n   The helpers can also be invoked by absolute path from any working directory.\n\n## Composition patterns\n\n### Dynamics and filtering\n\nSubclass `NLS`, implement pure `state_transition(state, input, t)` and\n`observation(state, input, t)`, and call the model with `(state, input)` for one\nstep. `NLS.forward` evaluates both functions at the current clock and its hook\nincrements the clock afterwards. Call `set_refpoint` before reading `A`, `B`,\n`C`, `D`, `c1`, or `c2`; this computes Jacobians at the selected state\u002Finput\u002Ftime.\nThen give the same model to one filter and call it with an estimate, a\nmeasurement, an input, and a covariance. The filter's direct model evaluations\nare not a substitute for adding process\u002Fmeasurement noise to the actual data.\n\nFor a linear problem, use `LTI` or a time-indexed `LTV` instead of making a\nlinear `NLS`; their matrices can be fed directly to LQR. See the exact time\nsemantics and observation timing in the references.\n\n### Control\n\nUse `LQR(system, Q, p, T)` for one finite-horizon solve. `Q` covers the\nconcatenated `[state, input]` vector and `p` has the same final feature size.\nUse `MPC` around the same system and costs when nonlinear dynamics require\nrelinearization. `MPC.forward` performs iterative LQR under `no_grad`, selects a\nbest iterate using its mutable stepper, and makes a final LQR call; set a small,\nexplicit stepper for tests and deterministic examples.\n\n### Sensor and pose branches\n\n`IMUPreintegrator` consumes body-frame angular rates\u002Faccelerations and positive\nfinite time intervals, returning incrementally propagated pose, velocity,\nposition, and optionally a 9x9 covariance. `EPnP` estimates an `SE3` from at\nleast four 3D\u002F2D correspondences and camera intrinsics. `ICP` estimates an `SE3`\nfrom source\u002Ftarget point clouds by nearest-neighbor matching and SVD; it is\nlocal and initialization-sensitive. Keep these frame conventions documented at\ncall sites instead of silently composing poses in an inverse order.\n\n## Verification gates\n\nBefore handing off code that uses this skill:\n\n- import and constructor signatures match [api-reference.md](references\u002Fapi-reference.md);\n- a minimal deterministic call succeeds from a clean current working directory;\n- all tensors have declared trailing shapes and common dtype\u002Fdevice;\n- filter `P`, `Q`, and `R` are square in their intended state\u002Fobservation spaces;\n- dynamics clock\u002Freset and IMU accumulation\u002Freset behavior are tested;\n- LQR\u002FICP\u002FMPC stepper limits are explicit for tests;\n- geometric inputs meet point-count, intrinsics, non-degeneracy, and\n  initialization requirements; and\n- no smoke helper downloads data, plots, writes outside an explicitly selected\n  output, or depends on the caller's working directory.\n\nThe references summarize the verified public module contracts and synthetic\nsmoke scripts; a Researcher does not need the original source checkout or\nexample files to use this sub-skill.\n","# PyPose robotics-modules\n\nUse this skill when a Researcher needs PyPose's high-level robotics modules in\n`pypose.module`: system models (`System`, `LTI`, `LTV`, `NLS`), EKF\u002FUKF\u002FPF,\nLQR\u002FMPC, `IMUPreintegrator`, `EPnP`, `ICP`, or `GeodesicLoss`. The module\nimplementations are differentiable PyTorch `nn.Module`s, but several modules\nalso mutate a clock, integration state, or stopping planner. Treat those state\nand shape contracts as part of the API.\n\n## Scope boundary\n\n- **Included:** constructors, forward signatures, rightmost feature dimensions,\n  leading-batch broadcasting, time and reset behavior, covariance contracts,\n  dynamics\u002Ffilter\u002Fcontrol composition, IMU units and frame conventions, PnP and\n  ICP initialization, and solver-stepper diagnostics.\n- **Excluded:** primitive LieTensor\u002Fmanifold algebra (`lie-tensor`), generic\n  Gauss--Newton\u002FLevenberg--Marquardt optimization (`optimization`), and\n  geometry\u002Fspline\u002Ftrajectory metrics (`geometry-evaluation`). `EPnP(refine=True)`\n  may call its own internal refinement; use the PnP contract here rather than\n  treating that implementation detail as a general optimizer workflow.\n\nRead the smallest relevant reference before writing code:\n\n- [api-reference.md](references\u002Fapi-reference.md) for verified signatures and\n  exact call\u002Freturn contracts.\n- [state-estimation.md](references\u002Fstate-estimation.md) for `System`\u002F`NLS`\n  models and EKF\u002FUKF\u002FPF covariance and shape rules.\n- [control-and-dynamics.md](references\u002Fcontrol-and-dynamics.md) for LTI\u002FLTV,\n  LQR, MPC, horizon layouts, and stepper behavior.\n- [imu-and-geometric-solvers.md](references\u002Fimu-and-geometric-solvers.md) for\n  IMU, EPnP, ICP, and geodesic rotation loss.\n- [troubleshooting.md](references\u002Ftroubleshooting.md) for failure diagnosis and\n  safe recovery.\n\n## Operating workflow\n\n1. **Identify the module boundary.** Decide whether the task needs a dynamic\n   model, a state estimator, a controller, sensor preintegration, a geometric\n   solver, or a rotation loss. Do not replace a module with primitive manifold\n   operations unless the task explicitly needs those primitives.\n2. **Normalize shapes before constructing modules.** Put features in the last\n   dimension, matrices in the last two dimensions, and batch\u002Ftime axes in\n   leading positions. Make `dtype` and `device` consistent across model,\n   covariance, costs, and observations. Do not rely on accidental broadcasting\n   for state versus time axes.\n3. **Build the dependency chain.** A normal filtering chain is\n   `NLS -> (EKF | UKF | PF)`; a control chain is `NLS -> set_refpoint -> LQR`\n   inside `MPC`; a sensor chain is `IMUPreintegrator -> pose\u002Fvelocity\u002Fposition`\n   consumers. `EPnP` and `ICP` each return an `SE3` pose for a separate\n   geometry-registration branch, while `GeodesicLoss` compares rotations.\n4. **Make mutable state explicit.** Reset a dynamic model before replaying a\n   trajectory. Reuse an IMU integrator only when its accumulation semantics are\n   intended. Reuse a `ReduceToBason`-like stepper only after its `reset()`.\n   Avoid concurrent calls through one mutable module.\n5. **Check outputs, not just execution.** Assert expected trailing shapes,\n   finiteness, covariance symmetry\u002Fpositive semidefiniteness within numerical\n   tolerance, plausible units, and an error or cost that improves on a\n   deterministic fixture. Use the bundled scripts for a quick CPU check:\n\n   From the `pypose` skill directory, run:\n\n   ```bash\n   python sub-skills\u002Frobotics-modules\u002Fscripts\u002Ffilter_smoke.py\n   python sub-skills\u002Frobotics-modules\u002Fscripts\u002Ficp_smoke.py\n   ```\n\n   The helpers can also be invoked by absolute path from any working directory.\n\n## Composition patterns\n\n### Dynamics and filtering\n\nSubclass `NLS`, implement pure `state_transition(state, input, t)` and\n`observation(state, input, t)`, and call the model with `(state, input)` for one\nstep. `NLS.forward` evaluates both functions at the current clock and its hook\nincrements the clock afterwards. Call `set_refpoint` before reading `A`, `B`,\n`C`, `D`, `c1`, or `c2`; this computes Jacobians at the selected state\u002Finput\u002Ftime.\nThen give the same model to one filter and call it with an estimate, a\nmeasurement, an input, and a covariance. The filter's direct model evaluations\nare not a substitute for adding process\u002Fmeasurement noise to the actual data.\n\nFor a linear problem, use `LTI` or a time-indexed `LTV` instead of making a\nlinear `NLS`; their matrices can be fed directly to LQR. See the exact time\nsemantics and observation timing in the references.\n\n### Control\n\nUse `LQR(system, Q, p, T)` for one finite-horizon solve. `Q` covers the\nconcatenated `[state, input]` vector and `p` has the same final feature size.\nUse `MPC` around the same system and costs when nonlinear dynamics require\nrelinearization. `MPC.forward` performs iterative LQR under `no_grad`, selects a\nbest iterate using its mutable stepper, and makes a final LQR call; set a small,\nexplicit stepper for tests and deterministic examples.\n\n### Sensor and pose branches\n\n`IMUPreintegrator` consumes body-frame angular rates\u002Faccelerations and positive\nfinite time intervals, returning incrementally propagated pose, velocity,\nposition, and optionally a 9x9 covariance. `EPnP` estimates an `SE3` from at\nleast four 3D\u002F2D correspondences and camera intrinsics. `ICP` estimates an `SE3`\nfrom source\u002Ftarget point clouds by nearest-neighbor matching and SVD; it is\nlocal and initialization-sensitive. Keep these frame conventions documented at\ncall sites instead of silently composing poses in an inverse order.\n\n## Verification gates\n\nBefore handing off code that uses this skill:\n\n- import and constructor signatures match [api-reference.md](references\u002Fapi-reference.md);\n- a minimal deterministic call succeeds from a clean current working directory;\n- all tensors have declared trailing shapes and common dtype\u002Fdevice;\n- filter `P`, `Q`, and `R` are square in their intended state\u002Fobservation spaces;\n- dynamics clock\u002Freset and IMU accumulation\u002Freset behavior are tested;\n- LQR\u002FICP\u002FMPC stepper limits are explicit for tests;\n- geometric inputs meet point-count, intrinsics, non-degeneracy, and\n  initialization requirements; and\n- no smoke helper downloads data, plots, writes outside an explicitly selected\n  output, or depends on the caller's working directory.\n\nThe references summarize the verified public module contracts and synthetic\nsmoke scripts; a Researcher does not need the original source checkout or\nexample files to use this sub-skill.\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\u002Fpypose\u002Fsub-skills\u002Frobotics-modules",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 robotics-modules",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill robotics-modules",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill robotics-modules","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002FSKILL.md",[47,53,57,61,65,69,75],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fapi-reference.md","REFERENCE","参考文档",7574,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Freferences\u002Fapi-reference.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Fcontrol-and-dynamics.md",5880,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Freferences\u002Fcontrol-and-dynamics.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Fimu-and-geometric-solvers.md",6806,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Freferences\u002Fimu-and-geometric-solvers.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fstate-estimation.md",7034,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Freferences\u002Fstate-estimation.md",{"path":66,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Ftroubleshooting.md",6020,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Freferences\u002Ftroubleshooting.md",{"path":70,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":73,"sourceHtmlUrl":74},"scripts\u002Ffilter_smoke.py","SCRIPT","脚本",3410,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Fscripts\u002Ffilter_smoke.py",{"path":76,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":77,"sourceHtmlUrl":78},"scripts\u002Ficp_smoke.py",1848,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpypose\u002Fsub-skills\u002Frobotics-modules\u002Fscripts\u002Ficp_smoke.py",[80,87,94,101,108,115],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"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":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"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":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"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":102,"name":103,"nameZh":6,"description":104,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":105,"categories":106,"keywords":107,"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":109,"name":110,"nameZh":6,"description":111,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":112,"categories":113,"keywords":114,"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":116,"name":117,"nameZh":6,"description":118,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":119,"categories":120,"keywords":121,"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",[],[]]