pillumina/ascend-sleuth / skills/diagnose
diagnose
Verified昇腾训练/推理问题的核心诊断循环:收集症状、按 triage-tree 路由、两阶段加载 并验证 Tier 2 case、命中给 fix(高危 root cause 改提示 halt)或转深度排查。 Tier-2 未命中但最终解决时起草候选 case。全程写 trace。 仅在能执行命令的 agent(Claude Code / Codex / pi)中可用。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 45.1 KB |
| references/collect-gates.yaml | config | 17.4 KB |
| references/diagnosis-procedure.md | markdown | 46.4 KB |
| references/diagnosis-trace.md | markdown | 46.2 KB |
| references/platform-dispatch.md | markdown | 1.6 KB |
| references/procedure-gates.yaml | config | 4.3 KB |
| references/report-template.md | markdown | 23.7 KB |
| references/script-integration.md | markdown | 2.2 KB |
| references/src-fetch.md | markdown | 9.2 KB |
references/procedure-gates.yaml
View on GitHub# 流程闸门表(diagnose 的「方法缺口」消费点)——结构化绑定,不写散文
#
# 为什么存在(三轮盲测 + 第四轮近邻对/负对照的结论):
# 流程类先验知识(type: methodology)的失效原因不是载体形态,而是**加载时机与深度**——
# 实测结论三条:
# ① 形态无关:同一份流程写成 flow[] YAML 或散文 skill,被执行的保真度一致(R = S = 满分);
# ② 摘要行无效:只给摘要行与不给等效(P3 0/2 vs 0/2)——决定性判据会被摘要截断
# (例:摘要写"同步比例 > 0.2 则存在慢卡",却漏掉"慢卡 = WTR 最小的卡"这条反直觉规则);
# 给全文才改变结论(2/2)。
# ③ 选择可靠:只给 id/title/summary 索引让 agent 自己挑 → 7/7 选对(第四轮近邻对 + 负对照再测:16/16,
# 其中含一对"两条都像"的启动失败流程,以及 4 次"应弃用"的负对照)。
# 现状缺口:12 条真实 trace 里精度/性能会话**加载流程 0 次**——触发点缺失,不是内容缺失。
# 第四轮补测(用于定位触发点是否真的补上了):真·候选未命中时 C 臂 3/4 走到本步并加载全文;
# 而"记录为 miss"的旧样本 0/6 加载——因为那些 issue 事后已各自沉淀了 case,前提不再成立
# (回放记录会腐烂,见 eval/flow/README.md 的"使用纪律")。
#
# **时机统一在步骤 5(候选全部未命中后)**,不在候选之前。理由按**预算**(原则九):候选命中时流程用不上,
# 提前加载只是白付一次注意力预算。第四轮做了对照(流程先行 vs 候选先行,18 格):**11/11 未致偏离命中 case**
# ⇒"早加载会锚定"**未获支持**,不得再当实测结论引用;第三轮 009/010 的误导发生在**流程不覆盖该样本**时,
# 归因是"错流程"(内容面)而非"时机"(顺序面)。
#
# 分工(形态归 SKILL,内容归词条):本表只决定「何时该取流程、取到后怎么用」;
# 具体是哪一条流程由**仓库根先验层**的流程选择器选出:先读 `references/_procedure-index.yaml`
# (只有 `category → 分片` 一张小表),再打开 `references/_procedure-index/<category>.yaml`
# 那片,按 title/summary 选一条(注意:本文件在 `skills/diagnose/references/`,选择器在仓库根
# `references/`——两处同名不同目录)。**分片在本轮 category 下不存在时读全部分片**,不要
# 空手进深度排查。
# **本表不写症状关键词分支**——关键词写在闸门里等于与 triage-tree 双源,必漂移。
# **默认一条**;前提与现场证据明确矛盾时可换一条(受"连续失败 <=2"约束,记 conflict)。
# 流程与证据冲突时以证据为准——流程是"这类问题怎么查"的参考,不是"这次就是这个"的判词。
#
# 校验(verify_references.py,随 CI):kind=procedure 必须给 trigger + load: full + selector,
# 且 selector 指向的文件必须存在;caveat_refs 必须存在且 active。
gates:
- id: precision-procedure
category: precision
kind: procedure
trigger: >-
precision 类问题、**所有 Tier 2 候选都未命中**、进入步骤 5 深度排查时——
需要"按什么方法定界根因",而不是继续取数、也不是继续查错误签名
load: full
selector: references/_procedure-index.yaml
caveat_refs:
- msprobe-tool-side-effect-fault-patterns # 采集行为本身可能让问题消失——先排除再下结论
- id: performance-procedure
category: performance
kind: procedure
trigger: >-
performance 类问题、**所有 Tier 2 候选都未命中**、进入步骤 5 深度排查时——
需要按判据(阈值、分流规则)定位瓶颈环节
load: full
selector: references/_procedure-index.yaml
- id: interrupt-procedure
category: interrupt
kind: procedure
trigger: >-
interrupt 类问题、**所有 Tier 2 候选都未命中**、进入步骤 5 深度排查时——
需要按某条排查路径分流,而非继续 grep 签名
load: full
selector: references/_procedure-index.yaml
caveat_refs:
- ascend-log-troubleshooting # 日志量级/裁剪纪律:只贴失败 rank + 栈尾,别灌全量
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |