Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/sisibeloved/cpython-optimize-skill/workflow-platform-differential-discovery-deepdivenpx skills add sisibeloved/cpython-optimize-skill --skill workflow-platform-differential-discovery-deepdivegit clone --depth 1 https://github.com/sisibeloved/cpython-optimize-skillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sisibeloved/cpython-optimize-skill/workflow-platform-differential-discovery-deepdive)<a href="https://agentmods.dev/skills/sisibeloved/cpython-optimize-skill/workflow-platform-differential-discovery-deepdive"><img src="https://agentmods.dev/badge/skills/sisibeloved/cpython-optimize-skill/workflow-platform-differential-discovery-deepdive.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00077 | $0.01153 |
| Opus 5 | $0.00039 | $0.00576 |
| Sonnet 5 | $0.00015 | $0.00231 |
| Haiku 4.5 | $0.00008 | $0.00115 |
Grade A, and why
workflow-platform-differential-discovery-deepdive scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Platform Differential Discovery Deepdive Workflow
定位
按用例深钻的端到端剧本。输入是单个用例（来自粗筛 workflow-platform-differential-discovery 的候选用例清单，或用户直指），产物是单用例证据表 + 备选优化项。与粗筛 workflow 是流水线关系但相互独立——用户也可跳过粗筛直接指一个用例深钻。
Agent 分派
| 阶段 | 对应用户流程步骤 | Agent | 技能 | Gate |
|---|---|---|---|---|
| 0 选例 | ① 先钻哪个用例 | cinderx-orchestrator |
读粗筛清单或用户直指 | 用例已锁定，实验轴齐全 |
| 1 环境确认 | — | cinderx-environment-verifier |
cinderx-env-validate |
双平台环境可复用 |
| E1 性能基线 | ② 总性能数据 | pyperformance-benchmark-analyst |
pyperformance-result-compare |
差距 > 噪声带，置信区间不含 0 |
| E2 用例画像 | ③ 用例内容 + autojit 函数形状 | cinderx-jit-analyst |
cinderx-hir-lir-analyze |
函数形状与 perf 热点吻合 |
| E3 HIR 分布 | ④ HIR 分布 | cinderx-jit-analyst |
cinderx-hir-dump、cinderx-hir-lir-analyze |
差异点定位到节点级 |
| E4 LIR + wall clock | ⑤ LIR 分析，含调用数与 wall clock | cinderx-jit-analyst |
cinderx-hir-lir-analyze |
wall clock 拆解与 LIR 段对齐 |
| E5 差异点 | ⑥ 差异点（bb 级或单条机器码级） | cinderx-jit-analyst |
cinderx-hir-lir-analyze |
双平台指令逐条对齐，差异序列可枚举 |
| E6 根因下钻 | ⑦ 拆解根因到 ISA/指令集/微架构/硬件 | cinderx-evidence-analyst |
cinderx-evidence-table、cinderx-isa-microarch-compare |
根因落到 ISA/微架构/硬件，证据闭环（含采样可用性探测） |
| E7 优化方向 | ⑧ 优化方向 | cinderx-evidence-analyst |
cinderx-evidence-table |
策略 ↔ 根因一一对应 |
| E8 穿刺（跑） | ⑨ 穿刺 | cinderx-orchestrator → pyperformance-candidate-runner |
cinderx-ab-run-slot、pyperformance-worker-run |
隔离满足、可重复 |
| E8 穿刺（判读） | ⑨ 穿刺数据可信度 | cinderx-evidence-analyst |
cinderx-evidence-table |
收益落噪声带外，证据表签发可信度 |
| E9 优化价值 | ⑩ 备选优化项 | cinderx-evidence-analyst |
cinderx-evidence-table、cinderx-optimization-report |
收益/成本均有数据支撑 |
Gate
E1–E5 是证据采集（贴事实），E6–E9 是收口判读（下判断），分界线在 E6。任一阶段 Gate 不满足，停在该阶段补证据，不往后走：
- evidence-analyst 在 E6 接手时拿到 E1–E5 证据，填入证据表前 5 段（或标记
evidence_gap要求补），再推进 E6–E9。 - evidence-analyst 在 E6 有权回退要求 E3–E5 补证据（如指令没对齐回到 E5）。
- E8 拆两半：穿刺跑分（orchestrator + candidate-runner）与穿刺判读（evidence-analyst）分离，判读权归 evidence-analyst。
有了足够深层的根因（E6），才不被表面原因迷惑；有了优化方向（E7），才进入穿刺（E8）；穿刺数据可信，才判定优化价值（E9），产出最终备选优化项。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 38 lines · 77 tokens per session scan A fddacc57433c
workflow-platform-differential-discovery-deepdive is a skill published in the GitHub repository sisibeloved/cpython-optimize-skill (2 stars, last pushed 7d ago), licensed MIT. It adds 77 tokens to every session and 1,153 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
trace
Evidence-driven tracing lane that orchestrates competing tracer hypotheses in Claude built-in team mode.
smiles-validation
Strict SMILES validation, structural comparison, and modification verification. Catches invalid LLM-generated molecules.
mflux-debugging
Debug MLX ports by comparing against a PyTorch/diffusers reference via exported tensors/images (export-then-compare).
social-physics-filter
Unified boundary enforcement, interpersonal diagnostic, and relational audit engine. Absorbs 40 psychology + 2 social protocols and all relationship case studies.
cnsplots
Create, revise, and troubleshoot publication-ready scientific plots in Python with cnsplots, including distribution, regression, heatmap, genomics, survival, set, flow, and multi-panel figures. Use when a user asks for cnsplots code, Cell/Nature/Science-style visualization, precise pixel-sized figures, statistical…
thinking-scientific-method
When a symptom has several plausible causes, rank falsifiable hypotheses and run the cheapest discriminating observation first; prefer least-assumptive survivors only after evidence fit.