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 skills add mindspore-ai/akg --skill pypto-pitfallsgit clone --depth 1 https://github.com/mindspore-ai/akgWrote 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/mindspore-ai/akg/pypto-pitfalls)<a href="https://agentmods.dev/skills/mindspore-ai/akg/pypto-pitfalls"><img src="https://agentmods.dev/badge/skills/mindspore-ai/akg/pypto-pitfalls/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mindspore-ai/akg/pypto-pitfalls"><img src="https://agentmods.dev/badge/skills/mindspore-ai/akg/pypto-pitfalls.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00018 | $0.02862 |
| Opus 5 | $0.00009 | $0.01431 |
| Sonnet 5 | $0.00004 | $0.00572 |
| Haiku 4.5 | $0.00002 | $0.00286 |
Grade A, and why
pypto-pitfalls 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 10d 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 — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PyPTO 常见陷阱
1. 运算符规则(最高频错误)
+ *:标量任意位置。- /:tensor 必须在左。函数调用:第一参数必须 Tensor。
1.0 + x # OK(__radd__)
1.0 - x # CRASH(__rsub__ 未实现)
1.0 / x # CRASH(__rtruediv__ 未实现)
pypto.add(1.0, x) # CRASH(函数调用标量在前)
# 1 - x 正确写法
x * (-1.0) + 1.0 # 推荐
# 标量之间用 Python 运算
neg_delta = -delta # OK(delta 是闭包 float)
2. clamp / min(x, d) 实现
pypto.clamp 不可用。min(x, d) 用双重取反:-max(-x, -d)。pypto.minimum(x, 0.0) 可用。
# min(abs_diff, delta) — delta 是闭包 float
neg_abs = pypto.mul(abs_diff, -1.0)
clipped = pypto.mul(pypto.maximum(neg_abs, -delta), -1.0) # = min(abs_diff, delta)
Huber Loss 完整模式(必须用 clamp,不能简化):
diff = predictions - targets
abs_diff = pypto.abs(diff)
neg_abs = pypto.mul(abs_diff, -1.0)
clipped = pypto.mul(pypto.maximum(neg_abs, -delta), -1.0) # min(|d|, delta)
half_sq = clipped * clipped * 0.5
loss = half_sq + abs_diff - clipped # 完整 Huber 公式
total = pypto.sum(loss, dim=0, keepdim=True)
output[:] = total / flat_size
3. 工厂函数
标量参数(eps、slope、margin 等)必须作为工厂函数参数通过闭包传入 kernel。
3D+2D matmul 时,forward 展平后传展平维度 nm=N*M 给工厂,不要分别传 N、M。
4. matmul K > 65535
用逐元素乘法 + pypto.sum(a * b_broadcast, dim=1) 替代。forward 中 B.reshape(1, -1) 使其可广播。
5. 距离度量必须 sqrt
sum(diff*diff) 是平方距离,不是 L2 距离。TripletMarginLoss 等必须 pypto.sqrt(sum_sq + eps)。
6. tile rank = tensor rank
set_vec_tile_shapes 参数个数必须等于被操作 tensor 的 rank。2D tensor 用 2D tile。
6.1 盲抄 tile 常量(尤其 16384)
示例里的 16384/8192 是经验候选,不是固定答案。必须按当前 shape 和归约维重算。
- 常见误用:输入
(128, 4096)却写set_vec_tile_shapes(1, 16384)。 - 更合理候选:
set_vec_tile_shapes(4, 4096)(归约轴不浪费,且 batch 并行更高)。
要点:
- 优先避免明显
tile[i] > shape[i]的“预算浪费”。 - 示例代码只能借结构,不能照抄 shape/tile 数字。
6.2 把“少分段”误读成“归约轴越大越快”
“连续搬运达阈值后再调归约轴”是二级目标,但不是“归约轴 tile 越大越快”。
- 常见误用:直接写
tile_hidden = hidden,追求归约轴一次覆盖。 - 典型后果:UB/OoOSchedule 报错(即使语义正确也无法编译)。
正确做法:
- 先满足
prod(tile_shape) <= 16384与auto_tiles <= 2048。 - 若出现 UB/OoOSchedule 报错,优先降档:
16384 -> 8192 -> 4096。 - 若
auto_tiles > 2048,优先改为 loop 分块,不要硬塞更激进 tile。 - 先让连续搬运达到约
1KB(经验阈值),再在达标候选里做归约轴甜点比较(常试16/32/64)。
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.
- 10d ago First seen · 219 lines · 18 tokens per session scan A c6a229c785c6
pypto-pitfalls is a skill published in the GitHub repository mindspore-ai/akg (259 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 18 tokens to every session and 2,862 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
pypto-pitfalls
A reference for PyPTO syntax rules, including where scalar values and tensors may appear in expressions and function calls.
pypto-loop-view
A troubleshooting note for PyPTO programs that use `pypto.view`, explaining that its shape must be fixed when the code is compiled.
matlab
Build, review, migrate, and safely plan MATLAB or GNU Octave numerical workflows, including arrays, tabular/time data, tests, projects, graphics, MAT files, and explicit Python interoperability.
pennylane
Hardware-agnostic quantum ML framework with automatic differentiation. Use when training quantum circuits via gradients, building hybrid quantum-classical models, or needing device portability across IBM/Google/Rigetti/IonQ. Best for variational algorithms (VQE, QAOA), quantum neural networks, and integration with…
dd-code-generation
Use pup CLI for immediate Datadog operations or generate code for integration into applications.
rocm-kernels
Provides guidance for writing and benchmarking optimized Triton kernels for AMD GPUs (MI355X, R9700) on ROCm, targeting HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers. Core kernels: RMSNorm, RoPE 3D, GEGLU, AdaLN. Includes XCD swizzle, autotune, diffusers integration patterns, and LTX-Video pipeline…