Name: andrebrov/avatar-take
Rating: 68 (1 reviews)
Author: andrebrov

andrebrov/avatar-take

Generate a synthetic talking-head take from a written script using a cloned avatar and voice, in place of recording — then hand it to broll-sourcing so the rest of the pipeline runs unchanged. Builds the avatar and voice clone from the user's own existing footage via the HeyGen v3 API, and burns non-removable AI disclosure into every output. Use when the user wants to skip recording, test an AI avatar version of a reel, produce a localised version of a take they already recorded, or asks about cloning their own likeness or voice. Only ever for a likeness the operator owns.

Quality

85%

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

71%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.

Highly actionable, executable content with strong validation checkpoints and a clear consent/disclosure spine. It is held back by time-sensitive dates embedded in the main flow and a large inlined API-reference section that would benefit from being split into a reference file.

Suggestions

Move the dated regulatory/deprecation facts (EU AI Act in-force date, v2 removal 2026-10-31, free-tier end Feb 2026) into a short 'Current rules / deprecations' section so they don't pad the main workflow.

Extract the numbered v3 API curl sequence ('The v3 API, as it actually behaves') into a references/ file (e.g. HEYGEN_V3.md) and link to it one level deep from SKILL.md.

Add a single short numbered 'Generate' sequence at the top (setup → build avatar/voice → consent → generate → verify disclosure survived) so the overall flow is linear rather than implied by section order.

Dimension	Reasoning	Score
Conciseness	Mostly dense and efficient with no basic-concept padding, but several time-sensitive dates (EU AI Act '2 Aug 2026', v2 removal '2026-10-31', free tier 'February 2026') sit in the main flow rather than a deprecated/old-patterns section, and some war-story narrative could be trimmed.	3 / 5
Actionability	Fully copy-paste-ready: exact ffmpeg and curl commands, complete v3 request bodies with correct discriminators, precise endpoints and poll calls covering the common generation path.	5 / 5
Workflow Clarity	Strong explicit validation checkpoints (consent gate, balance pre-flight, and the disclosure-survival loop that re-applies per format then 'exits non-zero rather than shipping'), but the top-level flow is topic-organised rather than a single linear numbered sequence.	4 / 5
Progressive Disclosure	The one bundle file (scripts/avatar_take.py) is referenced correctly, but a substantial block of v3 API reference detail is inlined in SKILL.md rather than split into a separate reference file, so content that could live one level deeper is inline.	3 / 5
	Total	15 / 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 strong, concrete, third-person description that names specific actions, gives explicit 'Use when' triggers, and bounds the skill to owned likenesses. Minor synonym coverage gaps in trigger terms keep it just off perfect.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Generate a synthetic talking-head take', 'Builds the avatar and voice clone', 'burns non-removable AI disclosure into every output' — covering the full pipeline comprehensively.	5 / 5
Completeness	Explicitly answers both 'what' (generate synthetic take, build avatar/voice, burn disclosure) and 'when' via a clear 'Use when the user wants to skip recording, test an AI avatar version... or asks about cloning' clause with concrete triggers.	5 / 5
Trigger Term Quality	Natural user phrases like 'skip recording', 'AI avatar version of a reel', 'localised version of a take', 'cloning their own likeness or voice' are well covered; a few common synonyms (e.g. 'deepfake', 'voice clone') are not surfaced.	4 / 5
Distinctiveness Conflict Risk	A clear niche (synthetic talking-head generation from owned likeness via HeyGen v3) with distinct triggers and an explicit 'Only ever for a likeness the operator owns' boundary, giving minimal conflict risk.	5 / 5
	Total	19 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Reviewed

2 days ago

Table of Contents

Discovery Implementation Validation