← Back to Skills Marketplace
Visual Gen Ai Language
by
Gal Davidi
· GitHub ↗
· v1.2.1
1550
Downloads
4
Stars
1
Active Installs
3
Versions
Install in OpenClaw
/install vgl
Description
Maximum control over AI image generation — write structured VGL (Visual Generation Language) JSON that explicitly controls every visual attribute. Define exa...
Usage Guidance
This skill appears to be a translator that outputs structured JSON for image generation and does not itself request credentials or install code — that part is coherent. The main red flag is an inconsistent JSON schema: the referenced schema marks many human-specific properties as universally required, while the SKILL.md describes them as conditional. That inconsistency can cause validation failures or force the generator to invent irrelevant values. Before installing/using: 1) test the skill with simple prompts to see the exact JSON it emits; 2) if you plan to execute generated VGL JSON via the bria-ai (or any) execution skill, ensure that execution skill requires and stores API credentials separately and review its permissions; 3) ask the author or maintainer to clarify/fix the schema requirements (make human-only fields optional) to avoid accidental inclusion of sensitive personal attributes; and 4) avoid sending real-person-identifying images or private data through the skill without verifying the downstream execution environment and privacy controls.
Capability Analysis
Type: OpenClaw Skill
Name: vgl
Version: 1.2.1
The OpenClaw AgentSkills skill bundle is designed to generate structured JSON (VGL) for image generation. The `SKILL.md` file provides detailed instructions for the AI agent on how to construct this JSON, including schema, rules, and examples. There are no instructions that attempt prompt injection to subvert the agent's purpose, nor any code snippets or commands indicating data exfiltration, malicious execution, persistence, or unauthorized network activity. The `curl` example demonstrates how a user would interact with an external API using the agent's output, which is a standard and benign use case.
Capability Assessment
Purpose & Capability
Name/description claim to convert natural-language image requests into deterministic VGL JSON for Bria/FIBO; SKILL.md and examples are consistent with that translator/templating purpose. The skill is instruction-only and does not request unrelated credentials or binaries.
Instruction Scope
SKILL.md instructs the agent to emit a single VGL JSON object and references the bria-ai skill for execution. However, the schema in references/schema-reference.md declares many object-level properties (pose, expression, clothing, number_of_objects, skin_tone_and_texture, etc.) as required for every object entry — contradicting the SKILL.md text which says those are only for human subjects or clusters. This mismatch makes the runtime instructions ambiguous and could force inclusion of irrelevant or fabricated data to satisfy validation.
Install Mechanism
Instruction-only skill with no install spec and no code files; nothing is written to disk or downloaded. Lowest-risk install posture.
Credentials
The skill declares no required environment variables, credentials, or config paths. It references a separate bria-ai skill for execution, but does not itself request keys or other secrets.
Persistence & Privilege
Does not request always:true, has no system config writes or other privileged persistence. Default autonomous invocation is allowed (platform default) but not combined with other red flags.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install vgl - After installation, invoke the skill by name or use
/vgl - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.2.1
v1.2.1 — Minor update
- Added .DS_Store file (no user-facing changes).
- No changes to core functionality or documentation, except updated SKILL.md metadata (author, license, version).
- Restated and clarified the skill’s mission: maximum deterministic control for AI image generation with VGL JSON for Bria FIBO models.
- No breaking changes or new features.
v0.0.1
No changes detected in this version.
- Initial release (0.0.1) of the vgl skill.
- Provides structured VGL (Visual Generation Language) JSON prompt guidelines for Bria's FIBO image generation models.
- Includes detailed schema, usage scenarios, and edit instruction standards for image generation, editing, inpainting, outpainting, and captioning.
v0.1.0
v0.1.0 – Initial Release
- Introduces structured prompt writing for Bria's FIBO models using VGL (Visual Generation Language) in strict JSON schema.
- Supports multiple image generation and editing modes: generation, edit, inpainting (masked edit), captioning, and prompt refinement.
- Defines detailed schema for describing image content, objects, environments, lighting, aesthetics, photographic characteristics, and style.
- Includes explicit instructions for editing and inpainting, with fidelity rules to preserve key image properties.
- Ensures deterministic, reproducible descriptions for use with Bria/FIBO’s AI image generation APIs.
Metadata
Frequently Asked Questions
What is Visual Gen Ai Language?
Maximum control over AI image generation — write structured VGL (Visual Generation Language) JSON that explicitly controls every visual attribute. Define exa... It is an AI Agent Skill for Claude Code / OpenClaw, with 1550 downloads so far.
How do I install Visual Gen Ai Language?
Run "/install vgl" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Visual Gen Ai Language free?
Yes, Visual Gen Ai Language is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Visual Gen Ai Language support?
Visual Gen Ai Language is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Visual Gen Ai Language?
It is built and maintained by Gal Davidi (@galbria); the current version is v1.2.1.
More Skills