create-creator-takes-h3

Generate an AI creator talking to camera, saying an approved script, as one continuous track — H3 Max reference-to-video through the GooseWorks fal-proxy (bills the Ads agent). Plans takes on line boundaries under H3's 15s cap, dry-runs a cost estimate, generates the first take alone so its voice can be locked and passed to every later take, joins takes with measured 0.10s dissolves, and moves each line's timing onto the words actually spoken. Use for any format with a generated creator speaking a script (split-screen, screen inserts, talking-head ads).

Quality

91%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

Quality

Content

88%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.

An exemplary operational skill body: fully executable commands, explicit validation gates around every paid step, and hard-won failure knowledge stated as rules. The residual gaps are minor — a couple of history narratives could be cut and the bundle's largest file is not surfaced in the navigation table.

Suggestions

Add media_proxy.py to the script table (e.g. what it does, cost, when the skill calls it) so the entire bundle is discoverable from SKILL.md.

Split the single-paragraph "Measured join acceptance" section into short labeled rules (required word files, estimated-join legacy path, error cases) — it is the densest passage in the file.

Compress the guard-history narratives in "This is now enforced, not advisory" and failure mode 2 to one sentence each; the rule and the check command carry the value.

Dimension	Reasoning	Score
Conciseness	Almost every token is non-obvious, project-specific operational knowledge (voice locking, measured dissolve values, rejection-ledger semantics) rather than concepts Claude already knows, but the historical postmortem narratives ("It was advisory once and the docstring carried 'a man in his late 20s ...'") and the single dense paragraph in "Measured join acceptance" could be trimmed.	4 / 5
Actionability	The Run block gives copy-paste-ready commands with real flags (--reseed t1, --split-at 6.3,14.5, --require-words), the character.json template shows the exact fields, and failure handling includes concrete exit codes, grep checks, and measured thresholds.	5 / 5
Workflow Clarity	The paid, batch pipeline is explicitly sequenced (script, still, t1 alone, listen, rest, transcribe each take, join, align) with validation checkpoints at every spend point (dry-run estimate, user approval before --go, watch every take end to end) and feedback loops for recovery (--reseed, fix the still not the seed, scene checklist).	5 / 5
Progressive Disclosure	The opening script table gives clear one-level-deep navigation and every referenced bundle file (make_character.py, plan_takes.py, run_takes.py, join_takes.py, align_beats.py, character-prompt.json) exists in scripts/, but media_proxy.py — the largest bundle file — is missing from the table, and substantial reference detail (join acceptance rules, rejection/ledger semantics) is inlined rather than split out.	4 / 5
	Total	18 / 20 Passed

Description

92%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.

A highly specific, third-person description that states concrete end-to-end capabilities and an explicit Use-for trigger clause. Its only gap is trigger-term breadth: a few natural synonyms for the creator/ads domain would round it out.

Suggestions

Add one or two natural synonyms to the Use-for clause (e.g. "AI spokesperson" or "UGC-style creator ad") so users phrasing the need differently still hit the trigger.

Dimension	Reasoning	Score
Specificity	The description lists six concrete pipeline actions ("Plans takes on line boundaries under H3's 15s cap, dry-runs a cost estimate, generates the first take alone so its voice can be locked... joins takes with measured 0.10s dissolves, and moves each line's timing onto the words actually spoken"), covering the workflow comprehensively from planning to final alignment.	5 / 5
Completeness	It explicitly answers both questions: the 'what' is the full creator-track pipeline, and the 'when' is a concrete trigger clause — "Use for any format with a generated creator speaking a script (split-screen, screen inserts, talking-head ads)".	5 / 5
Trigger Term Quality	Good natural phrases ("AI creator talking to camera", "generated creator speaking a script", "talking-head ads") but coverage stops short of the anchor-5 bar: no synonyms such as spokesperson/avatar/UGC-style and no file extensions or format names beyond the three listed.	4 / 5
Distinctiveness Conflict Risk	Naming "H3 Max reference-to-video through the GooseWorks fal-proxy" and the voice-locking take pipeline establishes a clear niche with distinct triggers; minimal overlap risk with generic video or image skills.	5 / 5
	Total	19 / 20 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: gooseworks-ai/goose-skills
Path: skills/ads/capabilities/create-creator-takes-h3/SKILL.md
Commit: cc3e518

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.