@adbutler/skills
v1.0.0
Published
Reusable AI-agent skills for AdButler ad management workflows — campaign launch, VAST video, retail media, reporting, targeting, programmatic, contracts, channels, and drafts
Readme
AdButler Skills
Reusable AI-agent skills for AdButler ad management workflows. Each skill teaches AI coding agents (Claude Code, Cursor, Copilot) how to perform a specific AdButler workflow end-to-end.
Skills
| Skill | Description | |-------|-------------| | Campaign Launch | Create and launch a display ad campaign — advertiser, campaign, creatives, zones, scheduling, and placements | | Retail Media Setup | Set up a retail media network with catalog zones, sponsored products, auction pricing, and bid floors | | Reporting Dashboard | Pull performance reports, forecasts, geo summaries, and manage saved report configurations | | VAST Video | Create VAST video ad campaigns with linear/non-linear media, companions, and tracking events | | Programmatic | Set up programmatic advertising with demand sources, demand endpoints, and PMP deals | | Targeting & Audiences | Configure targeting rules — geo, platform, day-parting, data keys, list targets, audiences, and postal codes | | Contract Management | Manage advertising contracts, templates, documents, payments, and signature requests | | Channel Management | Create channels to group zones for unified campaign delivery across multiple publishers | | Draft Campaigns | Stage campaigns, ad items, and placements in draft mode before publishing live |
Prerequisites
- An AdButler account with API access
- API key from AdButler Settings > API Keys
- Set the environment variable:
export ADBUTLER_API_KEY=your_key_here - Node.js 18+ (for running JavaScript templates)
How It Works
Each skill contains:
skills/{skill-name}/
skill.json # Machine-readable metadata
instructions.md # AI-agent-readable step-by-step instructions
examples/ # 3 real-world usage examples
templates/ # Working JavaScript code
{skill}.mjs # fetch()-based implementation
{skill}-mcp.mjs # MCP tool sequence for when MCP server is connectedFor AI Agents
The instructions.md in each skill is written for AI agents, not humans. It includes:
- Numbered steps with exact API endpoints and required fields
- MCP tool mapping table for when the AdButler MCP server is connected
- Decision trees for handling variations
- Validation checklists to verify success
- Common error table for troubleshooting
MCP Server Integration
If the AdButler MCP server is connected, agents can use MCP tools directly instead of making API calls. Each skill's instructions include a mapping table showing which MCP tool corresponds to each step.
Install the MCP server: npx -y @adbutler/mcp-server
Usage with AI Coding Agents
Claude Code
Add the skill instructions to your project context:
# Reference in your CLAUDE.md
echo "See /path/to/adbutler-skills/skills/adbutler-campaign-launch/instructions.md for campaign workflow" >> CLAUDE.mdOr point Claude Code at the skill directly:
Read /path/to/adbutler-skills/skills/adbutler-campaign-launch/instructions.md and follow it to launch a campaignCursor
Add to your .cursorrules:
When working with AdButler, reference the skills in /path/to/adbutler-skills/skills/ for workflows.Copilot
Add to your workspace instructions to reference the skill files for AdButler workflows.
Running Templates Directly
Each skill includes runnable JavaScript templates:
# Set your API key
export ADBUTLER_API_KEY=your_key_here
# Run a campaign launch
node skills/adbutler-campaign-launch/templates/campaign-launch.mjs
# Run retail media setup
node skills/adbutler-retail-media-setup/templates/retail-media-setup.mjs
# Run VAST video campaign
node skills/adbutler-vast-video/templates/vast-video.mjsShared Library
The lib/ directory contains shared utilities used by all skill templates:
adbutler-client.mjs— fetch()-based API client for AdButler v2errors.mjs— Error classes and field validation helpersvalidators.mjs— API key and input validation
API Reference
All skills use the AdButler API v2:
- Base URL:
https://api.adbutler.com/v2 - Auth:
Authorization: Basic {API_KEY}header - Format: JSON request/response bodies
License
MIT
