chengfeng-cut-talking-head

剪辑中文口播原素材：逐词转录、识别口误与重复、生成删词候选、在 Studio 审核后执行可靠物理剪切，并重建剪后字幕。用户说剪口播、处理口误、生成口播基础素材、继续剪口播，或确认卡回传 action=continue_cut / return_cut_review 时使用。不要用于单独安装、单独打开工作台、普通视频编辑或口播分镜成片。

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

85%

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 high-quality, executable multi-stage workflow with strong CAS/revision validation and feedback loops, appropriate one-level reference split, and concrete commands throughout. Its one weakness is conciseness: the §2 rationale and dated incident notes duplicate content already in semantic-deletion.md.

Suggestions

In §2, reduce the inline playback-order rationale and the 2026-07-26 '430→394' incident to a one-line pointer to semantic-deletion.md's '判断之前' section; the reference already carries the full argument.

Move the dated incident notes (2026-07-26) into the reference file or a 'known failure modes' note so the stage overview stays lean and free of inline time-sensitive annotations.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and assumes Claude's competence, but the §2 playback-order rationale and the 2026-07-26 '430→394' incident are restated both inline and in semantic-deletion.md, so some content could be tightened by leaning more on the reference. Inline dated incident notes also add tokens that are not strictly needed in the overview.	2 / 3
Actionability	Fully executable bash with real flags, scripts (ensure-runtime/running/studio, videocut-cli), concrete revision fields, and a copy-paste-ready JSON candidate schema — matches the 'fully executable, copy-paste ready' anchor.	3 / 3
Workflow Clarity	Fixed stage contract (preflight → readback → proposal → CAS → review binding → confirmation → execution → verification) with explicit validation checkpoints (revision matching, readBackVerified, noLongerCut, edit_list_changed_after_confirmation re-display) and error-recovery feedback loops for a destructive/batch operation.	3 / 3
Progressive Disclosure	SKILL.md acts as a stage-orchestration overview with well-signaled one-level-deep references (references/semantic-deletion.md and references/subtitle-correction.md both present in the bundle), and detail appropriately split out; navigation is easy and clearly signaled.	3 / 3
	Total	11 / 12 Passed

Description

100%

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 concise yet complete: concrete actions, natural triggers, an explicit 'Use when' clause, and clear exclusions that bound its scope. It is written in third person and reads as a tight, well-targeted trigger description.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "逐词转录、识别口误与重复、生成删词候选、…执行可靠物理剪切，并重建剪后字幕" — matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	Explicitly answers both what (the action list) and when ("…时使用") with explicit trigger guidance, and adds a negative-scope clause, so it clears the 'Use when…' bar rather than capping at 2.	3 / 3
Trigger Term Quality	Natural trigger phrases a user would say ("剪口播、处理口误、生成口播基础素材、继续剪口播") plus the card action tokens (action=continue_cut / return_cut_review) give good coverage of terms users would actually invoke.	3 / 3
Distinctiveness Conflict Risk	Niche is narrowly defined (Chinese talking-head cut editing) with distinct triggers and explicit exclusions ("不要用于单独安装、单独打开工作台、普通视频编辑或口播分镜成片"), making conflict with other skills unlikely.	3 / 3
	Total	12 / 12 Passed

Validation

93%

Warnings & errors only

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

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 2 suspicious	Warning

	Total	15 / 16 Passed

Repository: Agentchengfeng/chengfeng-videocut-skills
Path: plugins/chengfeng-videocut/skills/chengfeng-cut-talking-head/SKILL.md
Commit: 271e91b

Reviewed: about 13 hours ago

Table of Contents

Discovery Implementation Validation

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.