video-forge

视频制作全链路：素材入库 → 剧本冻结 → 全局配音 → 对齐 → 渲染 → 审查 → 交付。 Use when: 已明确要制作视频、做 showcase、做教程视频、录屏剪辑、video review、节奏审查。 Not for: Agent 产品宣传片仍未锁定主角、观众信念与导演语法（先用 agent-product-promo-director）；只找时效性外部参考（用 deep-research）；纯代码开发（用 worktree/tdd）、纯文档写作（直接写）、PPT（用 ppt-forge）。 Output: schema 驱动的视频成片 + operator 创意验收 + 风险匹配的技术审查 + 可发布。

Quality

73%

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

Fix and improve this skill with Tessl

tessl review fix ./cat-cafe-skills/video-forge/SKILL.md

Quality

Content

60%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 content is a well-structured, actionable production-line skill with explicit gates and feedback loops, but it is moderately verbose and—critically—references a refs/continuity.md bundle that does not exist, so progressive disclosure breaks. Tightening the repeated gate philosophy and adding the missing refs/continuity.md would lift both conciseness and progressive_disclosure.

Suggestions

Create the referenced refs/continuity.md (事前预测清单、降级路径、15s 失败案例) so the [refs/continuity.md] links resolve, or inline its essentials and drop the link.

Consolidate the repeated subject/motion philosophy spread across Intent Gate, Narrative Subject Gate, F2.5 Motion Evidence Gate, and the '常见错误' table into a single '铁规矩' section to cut redundancy.

Trim or relocate the three dated '失败史' callout blocks into a single 'lessons learned' reference so the main flow stays lean.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient and table-driven, but it repeats gate philosophy across Intent Gate / Narrative Subject Gate / Motion Evidence Gate / common-errors with overlapping prose, and the three '失败史' callout blocks plus '铁规矩' restate principles already in the gate tables. Not a 2 because content is dense and skill-specific rather than generic padding; not a 4 because several paragraphs could be tightened without losing clarity.	3 / 5
Actionability	Provides concrete executable commands (ffmpeg concat, CRF 23 compression, rich-block JSON snippets) and specific decision tables. Not a 5 because some guidance is policy/gate prose rather than copy-paste-ready commands, and the retiming/AI-generation steps are partly procedural rather than fully runnable examples; not a 3 because most key operations have real code.	4 / 5
Workflow Clarity	The 场景路由 table sequences A→G plus R (Patch Loop), with explicit gates (F1/F2/F2.5/F3) as validation checkpoints and BLOCK→degradation recovery loops. Not a 5 because some checkpoints (e.g. Seam Review 5-question gate) rely on qualitative answers without a hard re-run/re-validate command; not a 3 because validation gates and feedback loops are clearly present.	4 / 5
Progressive Disclosure	The body references [`refs/continuity.md`](refs/continuity.md) (twice) and ../../docs/features/F138-video-studio.md, but no refs/ directory or any references/scripts/assets bundle exists in the skill folder, so the referenced detailed material is absent/buried and navigation dead-ends. Not a 1 because there is section structure and a real (if missing) reference target; not a 3 because the one-level-deep reference it signals does not exist, breaking the disclosure contract.	2 / 5
	Total	13 / 20 Passed

Description

87%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 strong: it states a concrete pipeline, provides explicit 'Use when' triggers, and uniquely includes a 'Not for' boundary that disambiguates it from sibling skills. Minor weakness is that the capability list reads as pipeline phases rather than a granular action enumeration.

Dimension	Reasoning	Score
Specificity	Names a concrete end-to-end pipeline ('素材入库 → 剧本冻结 → 全局配音 → 对齐 → 渲染 → 审查 → 交付') with several specific actions, but stages are stage-names rather than the granular actions a 5 would enumerate. Not a 3 because coverage is broad and concrete; not a 5 because the actions are pipeline phases rather than a comprehensive enumeration of distinct operations.	4 / 5
Completeness	Explicitly answers 'what' (the full pipeline + 'Output: schema 驱动的视频成片...'), 'when' ('Use when: ...'), and even negative scope ('Not for: ...'), with concrete trigger phrases. This matches the 5 anchor that answers both what and when explicitly; no higher level exists.	5 / 5
Trigger Term Quality	The 'Use when' clause lists natural user phrases ('做 showcase、做教程视频、录屏剪辑、video review、节奏审查') a user would actually say, giving good keyword coverage. Not a 3 because multiple natural variations are present; not a 5 because a few obvious synonyms/extensions (e.g. 字幕, 配音, 剪辑 alone) are not all enumerated.	4 / 5
Distinctiveness Conflict Risk	A 'Not for' clause actively routes away from sibling skills (agent-product-promo-director, deep-research, worktree/tdd, ppt-forge), giving a clear niche with minimal conflict risk. Matches the 5 anchor's distinct-triggers benchmark.	5 / 5
	Total	18 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 3 missing, 1 suspicious	Warning

	Total	14 / 16 Passed

Repository: zts212653/clowder-ai
Path: cat-cafe-skills/video-forge/SKILL.md
Commit: 090626a

Reviewed: 1 day 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.