[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78SEZJPH9SBCMC2JTRRTXDD":3,"skill-same:01M78SEZJPH9SBCMC2JTRRTXDD":67},{"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},"01M78SEZJPH9SBCMC2JTRRTXDD","project-workflows",null,"Route Align-Anything satellite project workflows and decide runnable versus extension versus reference-only use.","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\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: project-workflows\ndescription: \"Route Align-Anything satellite project workflows and decide\n  runnable versus extension versus reference-only use.\"\nmetadata:\n  disco-role: operating\ndisable-model-invocation: true\nlicense: Apache 2.0\n---\n\n# Project Workflows\n\nUse this sub-skill when a task mentions an Align-Anything satellite project under `projects\u002F`, project-local scripts, Janus workflows, InterMT, language-feedback generation, text-image-to-text-image Chameleon workflows, any-to-text model initialization, or the bundled Eval-Anything project. This sub-skill helps decide whether the project material is immediately runnable, an extension pattern for core Align-Anything trainers, or reference-only evidence.\n\nDo not use this sub-skill as the primary guide for ordinary Align-Anything trainer modules, core package APIs, or generic installation. Route those through the root skill or the more specific training\u002Fevaluation sub-skills, then return here only for project-folder evidence.\n\n## First Decision\n\n1. Identify the project signal in the user request: `any_to_text`, `janus`, `intermt`, `lang_feedback`, `text_image_to_text_image`, or `eval-anything`.\n2. If working inside an Align-Anything checkout and the user asks what is available, run the bundled discovery script. It parses files and shell snippets but does not import or execute project code:\n\n   ```bash\n   python scripts\u002Flist_project_entrypoints.py --root \u003Crepository-root>\n   ```\n\n   Use `--json` for machine-readable output.\n3. Read `references\u002Fproject-map.md` for routing status, prerequisites, data-shape expectations, and runnable\u002Fextension\u002Freference-only decisions.\n4. For `eval-anything`, also read `references\u002Feval-anything-notes.md`; treat it as a separate package surface unless a dedicated runtime has been prepared.\n5. When something fails or appears inconsistent with the project README, read `references\u002Ftroubleshooting.md` before assuming the core package is broken.\n\n## Routing Matrix\n\n| Project signal | Default treatment | Runnable only when | Use as extension\u002Freference when |\n| --- | --- | --- | --- |\n| `projects\u002Fany_to_text` | Runnable builder scripts plus training-pattern evidence. | Align-Anything imports work, Transformers\u002FTorch can load the requested base LLM, CLIP vision tower, and optional CLAP audio tower, and the output directory is intentionally chosen. | Designing custom multimodal initializers or interpreting staged any-to-text training flags. |\n| `projects\u002Fjanus` and `scripts\u002Fjanus` | Optional Janus workflow. | The separate Janus-compatible package is installed, Janus model weights are available, GPU capacity is planned, and tokenized `.pt` data conventions match the target trainer. | Understanding how Align-Anything wires Janus SFT\u002FDPO generation and understanding trainer shell patterns. |\n| `projects\u002Fintermt` | Reference-only by default. | Only after a separate InterMT-Bench\u002Fdata runtime is intentionally prepared. | Explaining InterMT dataset\u002Fbenchmark intent and selecting multi-turn multimodal preference-alignment evidence. |\n| `projects\u002Flang_feedback` | Internal\u002Fdevelopment workflow; cautious runnable pattern. | vLLM, a multimodal model, GPU tensor parallel plan, images, and the expected JSON fields are available. | Reusing the base → critique → refine data-generation pattern without claiming stable public support. |\n| `projects\u002Ftext_image_to_text_image` | Chameleon preprocessing\u002Ftraining pattern with optional runtime. | A Chameleon-capable Transformers fork\u002Fmodel and GPU memory are available, and the dataset schema matches the selected tokenizer script. | Planning text-image interleaved SFT\u002FDPO\u002FRM\u002FPPO workflows, especially pre-tokenization before core trainers. |\n| `projects\u002Feval-anything` | Separate package\u002FCLI\u002Fpipeline reference unless a heavy runtime is prepared. | A Python 3.11-compatible Eval-Anything environment with vLLM\u002FHF\u002FAPI backend dependencies, model weights or API credentials, datasets, and GPU or backend resources is intentionally prepared. | Mapping safety benchmark configuration, package entry points, model backends, and pipeline extension points. |\n\n## Operating Rules\n\n- Treat project README commands as examples, not proof that a runtime is ready. Confirm optional packages, model identifiers, dataset schema, device count, and output directories first.\n- Do not run repository project scripts merely to discover capabilities. Use the bundled discovery script or inspect files statically.\n- Do not preserve placeholder path tokens in runnable commands; replace them with user-supplied locations or stop and ask.\n- Keep `eval-anything` separate from the main `align_anything` package. It has its own package metadata, CLI, configs, benchmark registry, model backend registry, and optional VLA dependency group.\n- If a project depends on another forked repository, external datasets, large model downloads, or API credentials, mark it optional or blocked until those are explicitly prepared.\n\n## Handoff Checklist\n\nWhen handing a project-workflow decision to another sub-skill or to execution, include:\n\n- Project folder and entrypoint name.\n- Decision: runnable, extension pattern, or reference-only.\n- Required package\u002Fruntime deltas beyond the base Align-Anything install.\n- Expected input schema and output artifact type.\n- Device\u002Fbackend assumptions and whether CPU substitution is unsupported, partial, or unknown.\n- Any known script caveat from `references\u002Ftroubleshooting.md`.\n","# Project Workflows\n\nUse this sub-skill when a task mentions an Align-Anything satellite project under `projects\u002F`, project-local scripts, Janus workflows, InterMT, language-feedback generation, text-image-to-text-image Chameleon workflows, any-to-text model initialization, or the bundled Eval-Anything project. This sub-skill helps decide whether the project material is immediately runnable, an extension pattern for core Align-Anything trainers, or reference-only evidence.\n\nDo not use this sub-skill as the primary guide for ordinary Align-Anything trainer modules, core package APIs, or generic installation. Route those through the root skill or the more specific training\u002Fevaluation sub-skills, then return here only for project-folder evidence.\n\n## First Decision\n\n1. Identify the project signal in the user request: `any_to_text`, `janus`, `intermt`, `lang_feedback`, `text_image_to_text_image`, or `eval-anything`.\n2. If working inside an Align-Anything checkout and the user asks what is available, run the bundled discovery script. It parses files and shell snippets but does not import or execute project code:\n\n   ```bash\n   python scripts\u002Flist_project_entrypoints.py --root \u003Crepository-root>\n   ```\n\n   Use `--json` for machine-readable output.\n3. Read `references\u002Fproject-map.md` for routing status, prerequisites, data-shape expectations, and runnable\u002Fextension\u002Freference-only decisions.\n4. For `eval-anything`, also read `references\u002Feval-anything-notes.md`; treat it as a separate package surface unless a dedicated runtime has been prepared.\n5. When something fails or appears inconsistent with the project README, read `references\u002Ftroubleshooting.md` before assuming the core package is broken.\n\n## Routing Matrix\n\n| Project signal | Default treatment | Runnable only when | Use as extension\u002Freference when |\n| --- | --- | --- | --- |\n| `projects\u002Fany_to_text` | Runnable builder scripts plus training-pattern evidence. | Align-Anything imports work, Transformers\u002FTorch can load the requested base LLM, CLIP vision tower, and optional CLAP audio tower, and the output directory is intentionally chosen. | Designing custom multimodal initializers or interpreting staged any-to-text training flags. |\n| `projects\u002Fjanus` and `scripts\u002Fjanus` | Optional Janus workflow. | The separate Janus-compatible package is installed, Janus model weights are available, GPU capacity is planned, and tokenized `.pt` data conventions match the target trainer. | Understanding how Align-Anything wires Janus SFT\u002FDPO generation and understanding trainer shell patterns. |\n| `projects\u002Fintermt` | Reference-only by default. | Only after a separate InterMT-Bench\u002Fdata runtime is intentionally prepared. | Explaining InterMT dataset\u002Fbenchmark intent and selecting multi-turn multimodal preference-alignment evidence. |\n| `projects\u002Flang_feedback` | Internal\u002Fdevelopment workflow; cautious runnable pattern. | vLLM, a multimodal model, GPU tensor parallel plan, images, and the expected JSON fields are available. | Reusing the base → critique → refine data-generation pattern without claiming stable public support. |\n| `projects\u002Ftext_image_to_text_image` | Chameleon preprocessing\u002Ftraining pattern with optional runtime. | A Chameleon-capable Transformers fork\u002Fmodel and GPU memory are available, and the dataset schema matches the selected tokenizer script. | Planning text-image interleaved SFT\u002FDPO\u002FRM\u002FPPO workflows, especially pre-tokenization before core trainers. |\n| `projects\u002Feval-anything` | Separate package\u002FCLI\u002Fpipeline reference unless a heavy runtime is prepared. | A Python 3.11-compatible Eval-Anything environment with vLLM\u002FHF\u002FAPI backend dependencies, model weights or API credentials, datasets, and GPU or backend resources is intentionally prepared. | Mapping safety benchmark configuration, package entry points, model backends, and pipeline extension points. |\n\n## Operating Rules\n\n- Treat project README commands as examples, not proof that a runtime is ready. Confirm optional packages, model identifiers, dataset schema, device count, and output directories first.\n- Do not run repository project scripts merely to discover capabilities. Use the bundled discovery script or inspect files statically.\n- Do not preserve placeholder path tokens in runnable commands; replace them with user-supplied locations or stop and ask.\n- Keep `eval-anything` separate from the main `align_anything` package. It has its own package metadata, CLI, configs, benchmark registry, model backend registry, and optional VLA dependency group.\n- If a project depends on another forked repository, external datasets, large model downloads, or API credentials, mark it optional or blocked until those are explicitly prepared.\n\n## Handoff Checklist\n\nWhen handing a project-workflow decision to another sub-skill or to execution, include:\n\n- Project folder and entrypoint name.\n- Decision: runnable, extension pattern, or reference-only.\n- Required package\u002Fruntime deltas beyond the base Align-Anything install.\n- Expected input schema and output artifact type.\n- Device\u002Fbackend assumptions and whether CPU substitution is unsupported, partial, or unknown.\n- Any known script caveat from `references\u002Ftroubleshooting.md`.\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\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows",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 project-workflows",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill project-workflows",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill project-workflows","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002FSKILL.md",[47,53,57,61],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Feval-anything-notes.md","REFERENCE","参考文档",6937,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002Freferences\u002Feval-anything-notes.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Fproject-map.md",14398,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002Freferences\u002Fproject-map.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Ftroubleshooting.md",10776,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002Freferences\u002Ftroubleshooting.md",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"scripts\u002Flist_project_entrypoints.py","SCRIPT","脚本",23780,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Falign-anything\u002Fsub-skills\u002Fproject-workflows\u002Fscripts\u002Flist_project_entrypoints.py",[68,75,82,89,96,103],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"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":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"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":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"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":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"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":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"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":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"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",[],[]]