workflow-routing-and-skill-catalog
Choose ARIS research workflows, slash skills, skill groups, common parameters, and artifact handoffs without mirroring every leaf skill body.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- workflow-routing-and-skill-catalog
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Choose ARIS research workflows, slash skills, skill groups, common parameters, and artifact handoffs without mirroring every leaf skill body.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/skill-groups.md2.5 KB
- 参考文档references/troubleshooting.md1.3 KB
- 参考文档references/workflow-map.md2.6 KB
03 SKILL.md 正文
Workflow Routing and Skill Catalog
Use this sub-skill when the user asks which ARIS skill to run, how W1-W6 connect, what effort or assurance means, how skill groups are selected, how Codex mirrors differ from mainline skills, or where ARIS artifacts flow between workflows.
Route Here
- Pick the right ARIS workflow for a research direction, experiment plan, review loop, paper draft, rebuttal, resubmission, talk, patent task, or utility task.
- Explain the common parameter axes:
effort,assurance, reviewer, GPU backend, human checkpoints, and auto-proceed behavior. - Diagnose why a workflow stopped because a required artifact is missing.
- Decide whether to invoke an orchestrator or a leaf skill.
- Maintain a user's minimal install set by mapping tasks to skill groups.
Reroute
- Installation mechanics, manifests, or skill-link failures:
../install-and-distribution/SKILL.md. - Reviewer/backend setup or cross-model review correctness:
../review-and-provider-backends/SKILL.md. - Research Wiki, session recovery, or watchdog state:
../state-recovery-and-experiment-ops/SKILL.md. - Source edits to skill catalog, mirrors, or tests:
../repository-maintenance/SKILL.md.
Fast Routing
- New research direction -> W1
/idea-discoveryor full/research-pipeline. - Existing experiment plan -> W1.5
/experiment-bridge. - Existing paper/results needing iterative critique -> W2
/auto-review-loopor a reviewer variant. - Narrative report ready for manuscript -> W3
/paper-writing. - Reviews received -> W4
/rebuttal. - Rejected paper to a new venue under no-new-experiment constraints -> W5
/resubmit-pipeline. - Accepted paper to presentation -> W6
/paper-talk. - Single utility need -> route directly to the leaf skill group instead of the whole orchestrator.
Reference Map
references/workflow-map.mdlists W1-W6, core artifacts, and when to use each path.references/skill-groups.mdsummarizes the catalog groups and representative skills.references/troubleshooting.mdcovers missing artifacts, over-broad workflows, group selection issues, and same-family review caveats.- Root
../../references/capability-map.mdmaps all major ARIS surfaces to sub-skills.
Important Distinctions
effortcontrols breadth/depth/budget;assurancecontrols audit strictness. They are independent.- At high assurance, audit gates may block a final report even if drafting succeeded.
- Codex base mirrors are useful for Codex execution but same-family review must be labeled provisional unless an independent gate is configured.
- HTML reports are readable views; Markdown/JSON artifacts remain canonical.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp 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.
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.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
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()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert 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.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match 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).