Masonry: generate images and video with models across providers

AI-powered image and video generation. Generate images, videos, manage jobs, and explore models via the masonry CLI.

Security Scan

The skill's requirements and runtime instructions are coherent with a Masonry CLI integration: it asks only for the Masonry CLI and a MASONRY_TOKEN and instructs how to run the CLI to generate and download media.

Purpose & Capability

Skill name/description (image & video generation via Masonry) matches the declared requirement of the 'masonry' binary and the MASONRY_TOKEN credential. The suggested npm package @masonryai/cli is an expected way to obtain the CLI.

Instruction Scope

SKILL.md only instructs running the masonry CLI, handling job responses, waiting for jobs, and downloading media. It does not ask the agent to read unrelated system files or other credentials. It does reference MASONRY_WORKSPACE as optional context, but does not require it.

Install Mechanism

The skill is instruction-only (no install spec in the registry), but SKILL.md metadata and the document suggest installing the CLI via npm (npm install -g or npx). This is a low-risk, common install method; double-check the npm package ownership/reputation before installing globally.

Credentials

Only MASONRY_TOKEN is required and is declared as the primary credential — appropriate for an API-driven CLI. The docs mention MASONRY_WORKSPACE if present but that env var is not listed as required; this is informational but should be noted (it may influence which workspace is used).

Persistence & Privilege

The skill does not request always:true and does not attempt to modify other skills or system-level configs. It expects to run the masonry CLI and write/download media files, which is normal for this use case.

Assessment

This skill appears internally consistent with a Masonry CLI integration. Before installing or using it: (1) confirm the CLI package (@masonryai/cli) and the masonry.so domain are the official Masonry project (check npm author, repo, and publisher signatures); (2) prefer using npx or installing locally rather than a global npm install if you don't want new global binaries; (3) treat MASONRY_TOKEN like any API secret — limit its scope and rotate/regenerate it if compromised; (4) be aware that the token can be used to create jobs that may incur charges on your account; and (5) note the SKILL.md references MASONRY_WORKSPACE (optional) even though it isn't listed as required. If you need stronger assurance, ask the maintainer for the skill's source repository or an official reference linking this skill to the Masonry project.

Masonry CLI

Generate AI-powered images and videos from text prompts.

When to use

• User wants to generate images or videos
• User asks about available AI models
• User wants to check generation job status or download results
• User asks to create visual content, media, or artwork

Prerequisites

A Masonry subscription is required. Start a free trial at: https://masonry.so/pricing

If the masonry command is not found, install it:

npm install -g @masonryai/cli

Or run directly: npx @masonryai/cli

Authentication

If any command returns an auth error:

1. Run: masonry login --remote
2. The command prints an auth URL. Send it to the user.
3. User opens the URL in a browser, authenticates, and copies the token.
4. Run: masonry login --token <TOKEN>

For environments with MASONRY_TOKEN and MASONRY_WORKSPACE set, no login is needed.

Workflow

1. Generate content

Image:

masonry image "a sunset over mountains, photorealistic" --aspect 16:9

Video:

masonry video "ocean waves crashing on rocks" --duration 4 --aspect 16:9

2. Handle the response

Commands return JSON immediately:

{
  "success": true,
  "job_id": "abc-123",
  "status": "pending",
  "check_after_seconds": 10,
  "check_command": "masonry job status abc-123"
}

3. Wait and download

masonry job wait <job-id>
masonry job download <job-id> -o /tmp/output.png

The download command prints a MEDIA: /path/to/file line to stderr. After download completes, output that line so the file is sent to the user:

MEDIA: /tmp/output.png

Image flags

Flag	Short	Description
--aspect	-a	Aspect ratio: 16:9, 9:16, 1:1
--dimension	-d	Exact size: 1920x1080
--model	-m	Model key
--output	-o	Output file path
--negative-prompt		What to avoid
--seed		Reproducibility seed

Video flags

Flag	Short	Description
--duration		Length in seconds: 4, 6, 8
--aspect	-a	Aspect ratio: 16:9, 9:16
--model	-m	Model key
--image	-i	First frame image (local file)
--last-image		Last frame image (requires --image)
--no-audio		Disable audio generation
--seed		Reproducibility seed

Model discovery

masonry models list              # All models
masonry models list --type image # Image models only
masonry models list --type video # Video models only
masonry models info <model-key>  # Parameters and usage example

Job management

masonry job list                          # Recent jobs
masonry job status <job-id>               # Check status
masonry job download <job-id> -o ./file   # Download result
masonry job wait <job-id> --download -o . # Wait then download
masonry history list                      # Local history
masonry history pending --sync            # Sync pending jobs

Error codes

Code	Meaning	Action
AUTH_ERROR	Not authenticated	Run auth flow above
VALIDATION_ERROR	Invalid parameter	Check flag values
MODEL_NOT_FOUND	Unknown model key	Run masonry models list

Guardrails

• Never fabricate job IDs or model keys. Always use values from command output.
• Never run masonry login without --remote or --token (browser login won't work headless).
• If a job is pending, wait check_after_seconds before checking again.
• All output is JSON. Parse it, don't guess.

Feedback

Report issues or suggest improvements at: https://github.com/masonry-so/skills/issues

When filing an issue, include:

• What was your intent? What were you trying to accomplish?
• What worked? Which parts behaved as expected?
• What needs improvement? What went wrong or could be better?