Agent Skill Layerproof

Layerproof API Skill Pack

⚠️ Skill Pack Manifest

This file is NOT a skill.

It is the entry point for the Layerproof skill pack.

Individual skills are located in:

skills/*/SKILL.md

AI agents and crawlers should only index those files as skills.

Overview

This repository provides a collection of AI agent skills for the Layerproof API.

These skills allow AI agents to interact with Layerproof to:

• Manage workspaces
• Create and manage projects
• Upload public and project files
• Generate slide decks using AI
• Edit slides and images
• Apply themes
• Export presentations
• Monitor asynchronous jobs

Each skill contains structured instructions that teach an AI agent how to interact with the Layerproof API using curl.

Required Environment Variables

Before performing any API operations, the agent must verify that these environment variables are set on the user's machine.

Check before every API call

Always verify the following environment variables exist before making any API request:

# Check if LAYERPROOF_BASE_URL is set
if [[ -z "${LAYERPROOF_BASE_URL}" ]]; then
  echo "ERROR: LAYERPROOF_BASE_URL is not set."
  echo "Please set it in your shell profile or .env file, then re-run."
  return 1
fi

# Check if LAYERPROOF_API_KEY is set
if [[ -z "${LAYERPROOF_API_KEY}" ]]; then
  echo "ERROR: LAYERPROOF_API_KEY is not set."
  echo "Please set it in your shell profile or .env file, then re-run."
  return 1
fi

Setting environment variables

If not already set, instruct the user to add them:

echo 'export LAYERPROOF_BASE_URL=https://api.layerproof.com' >> ~/.zshrc
echo 'export LAYERPROOF_API_KEY=your-api-key-here' >> ~/.zshrc
source ~/.zshrc

Alternatively, for project-specific settings, create a .env.local file in the project root:

LAYERPROOF_BASE_URL=https://api.layerproof.com
LAYERPROOF_API_KEY=your-api-key-here

When using a .env.local file, ensure it is loaded before running any curl commands:

# For zsh, load .env.local if it exists
if [[ -f .env.local ]]; then
  set -a
  source .env.local
  set +a
fi

Authentication header

All API requests must include the following header:

X-API-KEY: $LAYERPROOF_API_KEY

How Agents Should Use This Skill Pack

When a user asks to perform an operation related to Layerproof:

1. Identify the resource type involved.
2. Load the corresponding skills/<resource>/SKILL.md.
3. Follow the endpoint definitions in that skill file.
4. Construct a valid curl request.
5. Execute the request using the environment variables.
6. Display the raw JSON response to the user.

If the response contains image URLs, the agent may render them.

Skill Modules

Each module corresponds to a Layerproof API resource.

Handling Asynchronous Operations

Many Layerproof API endpoints run asynchronously.

When an API response returns an activityId, the agent should:

1. Poll the job status:

GET /api/v2/jobs/{activityId}

2. Continue polling until:

status = DONE

3. If failureReason exists, report the error.

Typical Agent Workflow

Example workflows (the agent should automatically select the correct skill module for each step).

Workflow A: Full presentation from scratch (with reference files and theme)

1. Workspaces — List workspaces (GET), then create one if needed (POST) or resolve workspace_id from list.
2. Projects — Create project (POST) with name, project_kind: "SLIDE_DECK", optional workspace_id; capture id and slide_deck_id.
3. Public files (optional) — If the user provides reference docs: POST /api/v2/files/prepare → user PUTs file to upload_url → POST /api/v2/files/confirm with s3_key; collect s3_key(s) for outline.
4. Themes (optional) — List themes (GET); if user wants a new look: POST /api/v2/themes/generate with prompt → poll Jobs until DONE → note theme_id.
5. Slide decks — POST .../outline/generate with prompt, slide_count, optional file_s3_keys, language; capture activity_id.
6. Jobs — Poll GET /api/v2/jobs/{activity_id} until status is DONE (or CANCELED); on failure, report failure_reason.
7. Slide decks — GET deck to read outline; optionally PUT .../outline to tweak sections (title, sections, key_points, visual_suggestion).
8. Themes / slide decks — If a theme was generated, POST /api/v2/themes/apply with slide_deck_id and theme_id (optional regenerate_slides). Then POST .../slides/batch-generate (optionally with generation_type, aspect_ratio); capture activity_id.
9. Jobs — Poll GET /api/v2/jobs/{activity_id} until DONE.
10. Slide decks — GET deck again; if a single slide needs regeneration, POST .../slides/{sectionId}/generate-transcript or .../generate-image etc., then poll job.
11. Slides (optional) — For a specific slide: POST image-edit or object-removal → poll job → POST accept-image-edit with live_object_id.
12. Exports — POST .../projects/{projectId}/exports/pptx, .../exports/png, or .../exports/video; capture export_id; poll GET .../exports/{export_id} until COMPLETED; present download_url to user.

Workflow B: Multi-format export and error handling

1. Projects — GET project by ID (or list and pick) to ensure it exists and has slide_deck_id.
2. Slide decks — GET deck; confirm metadata.completed_slides meets expectations before export.
3. Exports — Start PNG export (POST .../exports/png); get exportId. In parallel or after, start PPTX export (POST .../exports/pptx); get second exportId.
4. Exports — Poll each GET .../exports/{exportId} until status is COMPLETED or FAILED. If FAILED, show errorMessage and suggest retry or check project. When COMPLETED, show downloadUrl and expiresAt.
5. If user cancels: POST .../exports/{exportId}/cancel for in-progress exports.

Workflow C: Project files and outline with project-scoped assets

1. Projects — Create or get project; note project_id. Get or create a directory in the project (per project-files API if list directories exists) or use a known directory_id.
2. Project files — POST prepare with path, file_name, mime_type, size → user uploads to upload_url → POST confirm with file_id. Optionally GET file or download-url to verify.
3. Slide decks — Use project’s slide_deck_id; generate outline with file_s3_keys from project file S3 keys (if API accepts them) or use public files flow and pass file_s3_keys from public prepare/confirm.
4. Continue with outline → poll job → update outline → batch generate → poll → export as in Workflow A.

The agent should automatically select the correct skill module for each step and handle polling, error responses, and multi-step dependencies.

Skill Discovery

Agents should load the following files when interacting with Layerproof:

• skills/workspaces/SKILL.md
• skills/projects/SKILL.md
• skills/public-files/SKILL.md
• skills/project-files/SKILL.md
• skills/slide-decks/SKILL.md
• skills/slides/SKILL.md
• skills/themes/SKILL.md
• skills/tones/SKILL.md
• skills/exports/SKILL.md
• skills/social-campaigns/SKILL.md
• skills/jobs/SKILL.md

Each skill file contains:

• API endpoint definitions
• request and response types
• curl examples
• agent decision logic
• response handling rules

Repository Structure

agent-skill-layerproof
│
├─ SKILL.md
├─ marketplace.json
├─ README.md
│
└─ skills
   ├─ workspaces
   │   └─ SKILL.md
   ├─ projects
   │   └─ SKILL.md
   ├─ public-files
   │   └─ SKILL.md
   ├─ project-files
   │   └─ SKILL.md
   ├─ slide-decks
   │   └─ SKILL.md
   ├─ slides
   │   └─ SKILL.md
   ├─ themes
   │   └─ SKILL.md
   ├─ tones
   │   └─ SKILL.md
   ├─ exports
   │   └─ SKILL.md
   ├─ social-campaigns
   │   └─ SKILL.md
   └─ jobs
       └─ SKILL.md