Email deliveries in Adobe Campaign are stored in the database — inaccessible to external tools and invisible to AI agents without extraction.

Use acc instance pull to download all delivery HTML and text files locally, then run any AI agent directly in the folder to review copy, generate variants, or audit tone of voice at scale.

Preparation

Requirements: acc installed and authenticated. Refer to Getting started.

What will be downloaded:

  • Email deliveries as .html and .txt files, named by internal name

Walkthrough

1. Pull all email deliveries

acc instance pull --metadata nms:delivery

Deliveries are extracted to nms/delivery/ — one .html and one .txt per delivery.

2. Open the folder in an AI agent

Open nms/delivery/ in your AI agent of choice — Claude, Cursor, or GitHub Copilot.

3. Prompt with context

Give the agent brand and objective context before any task:

This folder contains recent email deliveries sent by [Company] via Adobe Campaign Classic.

Review DM_SPRING_2026.html and suggest 5 subject line variants for an A/B test.
Base the tone on the existing .txt files in this folder.

Adapt the prompt to your task: copy review, variant generation, tone-of-voice audit, or translation.

Validation & troubleshooting

The AI has no brand context

Add a BRAND.md or strategy document to the delivery folder before running the agent. The AI will use it as reference for tone and messaging.

Too many deliveries pulled

Use the schemas filter in acc.config.json to scope the pull by folder or label. Refer to Configuration reference.

Going further

The same approach works for personalization blocks and JavaScript codes — pull with acc, open in an AI agent, iterate locally. Refer to Decompose metadata into source files.