CtrlK
BlogDocsLog inGet started
Tessl Logo

deepep-to-cam-converter

识别代码中可替换为CAM算子的DeepEP算子,基于实际运行参数校验约束,自动完成通信域转换(NCCL->HCCL)、设备适配(CUDA->NPU)及算子替换。

64

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./jiuwenclaw/jiuwenclaw/resources/agent/jiuwenclaw_workspace/skills/deepep-to-cam-converter/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

85%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-structured, actionable migration workflow with explicit validation checkpoints, feedback loops, a final checklist, and clean progressive disclosure to five real reference files. The main improvement would be inlining the deep_ep->CAM interface mapping rather than deferring it entirely to references.

Suggestions

Inline a brief dispatch/combine interface-mapping example (deep_ep call -> corresponding CAM call) so the core replacement is executable without opening reference files.

De-duplicate the default-vs-runtime-value rule — state it once prominently and reference it from the checklist instead of repeating the full warning.

Convert the reference paths in '参考文档库' to markdown links (e.g. [name](references/...)) for clearer navigation.

DimensionReasoningScore

Conciseness

Mostly efficient and task-focused with no padding of concepts Claude already knows, though the default-vs-runtime-value rule is repeated across the overview, section 2.2, and the checklist, and some table cells are slightly verbose.

4 / 5

Actionability

Provides concrete, executable guidance — exact code patterns to find, replacement tables, specific import ordering, and exact comment strings — but defers the actual dispatch/combine interface mapping and parameter adjustments to reference docs rather than showing them inline.

4 / 5

Workflow Clarity

Four clearly sequenced phases each with explicit blocking self-check checkpoints, structured pass/fail output templates, feedback loops (fail -> stop -> re-execute), and a final verification checklist — validation is thorough for this destructive code-modification task.

5 / 5

Progressive Disclosure

SKILL.md serves as a clear overview pointing to five one-level-deep reference docs (all verified present in references/), each with a one-line description and path, with workflow and decision logic appropriately inlined.

5 / 5

Total

18

/

20

Passed

Description

67%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific and highly distinctive about the DeepEP-to-CAM migration task, but it omits any explicit 'when to use' trigger guidance, which caps completeness. Adding a 'Use when migrating DeepEP/MoE code to Ascend NPU...' clause would raise both completeness and trigger-term quality.

Suggestions

Add an explicit 'Use when...' trigger clause, e.g. 'Use when migrating DeepEP-based MoE code to Ascend (NPU) environments or replacing NCCL/CUDA with HCCL/NPU.'

Include natural user-facing phrases and synonyms ('昇腾', 'Ascend', 'MoE迁移', 'EP算子替换') so the description matches how users actually ask for this skill.

Consider mentioning file/code signals users would reference (e.g. 'import deep_ep', dispatch/combine calls) to improve trigger-term coverage.

DimensionReasoningScore

Specificity

Lists multiple concrete actions — identify DeepEP operators, validate constraints, convert comm domain (NCCL->HCCL), adapt devices (CUDA->NPU), and replace operators — giving comprehensive coverage of the migration task.

5 / 5

Completeness

Clearly answers 'what' but provides no explicit 'when to use' guidance; per the rubric, a missing 'Use when' clause caps completeness at 3.

3 / 5

Trigger Term Quality

Contains relevant domain keywords a user would mention (DeepEP, CAM, NCCL, HCCL, CUDA, NPU) but lacks natural trigger phrasing, synonyms, file extensions, and a 'Use when' clause.

3 / 5

Distinctiveness Conflict Risk

A highly specific niche (DeepEP->CAM, NCCL->HCCL, CUDA->NPU on Ascend NPU) with distinct triggers and minimal risk of firing for unrelated skills.

5 / 5

Total

16

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
XiaoLuoLYG/GOD
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.