[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7953HS5GHAJ387C9FDC5138":3,"skill-same:01M7953HS5GHAJ387C9FDC5138":60},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M7953HS5GHAJ387C9FDC5138","multi-agent-and-runner",null,"Route skrl 2.1.0 IPPO\u002FMAPPO multi-agent workflows, framework Runner configuration, model and memory dictionaries, trainer scopes, and distributed\u002Fcheckpoint boundaries.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"MIT","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: multi-agent-and-runner\ndescription: \"Route skrl 2.1.0 IPPO\u002FMAPPO multi-agent workflows, framework\n  Runner configuration, model and memory dictionaries, trainer scopes, and\n  distributed\u002Fcheckpoint boundaries.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# Multi-agent and Runner\n\nUse this route when the request names **IPPO**, **MAPPO**, PettingZoo or\nanother multi-agent environment, per-agent model\u002Fmemory dictionaries, a\ncentralized state\u002Fvalue function, Runner\u002FYAML configuration, trainer scopes,\nor multi-process\u002Fdistributed execution. It owns the multi-agent contract and\ndeclarative Runner machinery; it does not replace environment wrapping or the\nsingle-agent framework guides.\n\n## Route quickly\n\n1. Establish a framework-wrapped multi-agent environment first. Confirm\n   `possible_agents`, current `agents`, `observation_spaces`, `state_spaces`,\n   `action_spaces`, `device`, and `num_envs`. For PettingZoo or Isaac Lab,\n   start with [`..\u002Fenvironment-integration\u002FSKILL.md`](..\u002Fenvironment-integration\u002FSKILL.md).\n2. Choose the learning semantics in\n   [`references\u002Fmulti-agent-workflows.md`](references\u002Fmulti-agent-workflows.md):\n   IPPO is decentralized training\u002Fexecution with an independent value model;\n   MAPPO is centralized training\u002Fdecentralized execution through a state-based\n   value model. Both expose the same per-agent constructor shape.\n3. Build **nested dictionaries keyed by every `possible_agents` entry**:\n   `models[agent_id][\"policy\"]`, `models[agent_id][\"value\"]`, and, for\n   training, `memories[agent_id]`. A single-agent model dictionary is not a\n   multi-agent dictionary. Validate role keys, spaces, and model initialization\n   before constructing the algorithm.\n4. For declarative setup, use\n   [`references\u002Frunner-configuration.md`](references\u002Frunner-configuration.md).\n   A Runner needs a wrapped environment and top-level `models`, `memory`,\n   `agent`, and `trainer` configuration. Treat YAML as trusted input: Runner\n   evaluates selected component strings and expression fields.\n5. Select a trainer deliberately. A single IPPO\u002FMAPPO object normally uses\n   `SequentialTrainer`, `StepTrainer`, or direct loop ownership. `ParallelTrainer`\n   is a Torch multiprocessing facility, not the same thing as a multi-agent\n   environment. Scope validation and simultaneous-agent limitations are in the\n   workflow reference.\n6. Before resuming, check framework-specific checkpoint format, experiment\n   directory, rank ownership of logging, and model-role compatibility. Use\n   [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for shape,\n   key, migration, YAML, scope, and distributed failures.\n\n## Framework boundaries\n\n- **Torch:** IPPO and MAPPO are available; Torch additionally supports a\n  policy\u002Fvalue shared model through `shared_model` and automatic mixed\n  precision. `ParallelTrainer` is Torch-only.\n- **JAX:** IPPO and MAPPO are available with separate policy\u002Fvalue models.\n  Runner does not support shared policy\u002Fvalue models and removes the\n  unsupported mixed-precision setting. Call `init_state_dict` for generated\n  models before use.\n- **Warp:** the Warp Runner and sequential trainer cover single-agent Warp\n  agents. This release has no Warp IPPO\u002FMAPPO mapping; do not route a Warp\n  multi-agent request to this path as if it were supported. See\n  [`..\u002Fwarp-agent-training\u002FSKILL.md`](..\u002Fwarp-agent-training\u002FSKILL.md) for the\n  single-agent boundary.\n- Framework-specific model construction belongs to\n  [`..\u002Ftorch-agent-training\u002FSKILL.md`](..\u002Ftorch-agent-training\u002FSKILL.md),\n  [`..\u002Fjax-agent-training\u002FSKILL.md`](..\u002Fjax-agent-training\u002FSKILL.md), or\n  [`..\u002Fwarp-agent-training\u002FSKILL.md`](..\u002Fwarp-agent-training\u002FSKILL.md). Link\n  there for mixin or device details rather than duplicating those guides.\n\n## Completion checklist\n\nA safe handoff has: a wrapper-derived, immutable `possible_agents` list; all\nmodel and memory dictionaries covering it; policy\u002Faction and value\u002Fstate space\ncompatibility; an explicit choice between IPPO and MAPPO; a finite trainer\n`timestep` plan; validated scopes if multiple agents share a vectorized\nenvironment; trusted Runner\u002FYAML component names if used; explicit logging and\ncheckpoint paths; and a framework\u002Fdistributed boundary statement. Do not run\nfull multi-agent training or an unreviewed user YAML merely to diagnose its\nshape.\n","# Multi-agent and Runner\n\nUse this route when the request names **IPPO**, **MAPPO**, PettingZoo or\nanother multi-agent environment, per-agent model\u002Fmemory dictionaries, a\ncentralized state\u002Fvalue function, Runner\u002FYAML configuration, trainer scopes,\nor multi-process\u002Fdistributed execution. It owns the multi-agent contract and\ndeclarative Runner machinery; it does not replace environment wrapping or the\nsingle-agent framework guides.\n\n## Route quickly\n\n1. Establish a framework-wrapped multi-agent environment first. Confirm\n   `possible_agents`, current `agents`, `observation_spaces`, `state_spaces`,\n   `action_spaces`, `device`, and `num_envs`. For PettingZoo or Isaac Lab,\n   start with [`..\u002Fenvironment-integration\u002FSKILL.md`](..\u002Fenvironment-integration\u002FSKILL.md).\n2. Choose the learning semantics in\n   [`references\u002Fmulti-agent-workflows.md`](references\u002Fmulti-agent-workflows.md):\n   IPPO is decentralized training\u002Fexecution with an independent value model;\n   MAPPO is centralized training\u002Fdecentralized execution through a state-based\n   value model. Both expose the same per-agent constructor shape.\n3. Build **nested dictionaries keyed by every `possible_agents` entry**:\n   `models[agent_id][\"policy\"]`, `models[agent_id][\"value\"]`, and, for\n   training, `memories[agent_id]`. A single-agent model dictionary is not a\n   multi-agent dictionary. Validate role keys, spaces, and model initialization\n   before constructing the algorithm.\n4. For declarative setup, use\n   [`references\u002Frunner-configuration.md`](references\u002Frunner-configuration.md).\n   A Runner needs a wrapped environment and top-level `models`, `memory`,\n   `agent`, and `trainer` configuration. Treat YAML as trusted input: Runner\n   evaluates selected component strings and expression fields.\n5. Select a trainer deliberately. A single IPPO\u002FMAPPO object normally uses\n   `SequentialTrainer`, `StepTrainer`, or direct loop ownership. `ParallelTrainer`\n   is a Torch multiprocessing facility, not the same thing as a multi-agent\n   environment. Scope validation and simultaneous-agent limitations are in the\n   workflow reference.\n6. Before resuming, check framework-specific checkpoint format, experiment\n   directory, rank ownership of logging, and model-role compatibility. Use\n   [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for shape,\n   key, migration, YAML, scope, and distributed failures.\n\n## Framework boundaries\n\n- **Torch:** IPPO and MAPPO are available; Torch additionally supports a\n  policy\u002Fvalue shared model through `shared_model` and automatic mixed\n  precision. `ParallelTrainer` is Torch-only.\n- **JAX:** IPPO and MAPPO are available with separate policy\u002Fvalue models.\n  Runner does not support shared policy\u002Fvalue models and removes the\n  unsupported mixed-precision setting. Call `init_state_dict` for generated\n  models before use.\n- **Warp:** the Warp Runner and sequential trainer cover single-agent Warp\n  agents. This release has no Warp IPPO\u002FMAPPO mapping; do not route a Warp\n  multi-agent request to this path as if it were supported. See\n  [`..\u002Fwarp-agent-training\u002FSKILL.md`](..\u002Fwarp-agent-training\u002FSKILL.md) for the\n  single-agent boundary.\n- Framework-specific model construction belongs to\n  [`..\u002Ftorch-agent-training\u002FSKILL.md`](..\u002Ftorch-agent-training\u002FSKILL.md),\n  [`..\u002Fjax-agent-training\u002FSKILL.md`](..\u002Fjax-agent-training\u002FSKILL.md), or\n  [`..\u002Fwarp-agent-training\u002FSKILL.md`](..\u002Fwarp-agent-training\u002FSKILL.md). Link\n  there for mixin or device details rather than duplicating those guides.\n\n## Completion checklist\n\nA safe handoff has: a wrapper-derived, immutable `possible_agents` list; all\nmodel and memory dictionaries covering it; policy\u002Faction and value\u002Fstate space\ncompatibility; an explicit choice between IPPO and MAPPO; a finite trainer\n`timestep` plan; validated scopes if multiple agents share a vectorized\nenvironment; trusted Runner\u002FYAML component names if used; explicit logging and\ncheckpoint paths; and a framework\u002Fdistributed boundary statement. Do not run\nfull multi-agent training or an unreviewed user YAML merely to diagnose its\nshape.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill multi-agent-and-runner",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill multi-agent-and-runner",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill multi-agent-and-runner","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner\u002FSKILL.md",[46,52,56],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fmulti-agent-workflows.md","REFERENCE","参考文档",11378,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner\u002Freferences\u002Fmulti-agent-workflows.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Frunner-configuration.md",11166,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner\u002Freferences\u002Frunner-configuration.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Ftroubleshooting.md",10659,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fskrl\u002Fsub-skills\u002Fmulti-agent-and-runner\u002Freferences\u002Ftroubleshooting.md",[61,69,76,83,90,97],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"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.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"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":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"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":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"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":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"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":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":65,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"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",[],[]]