Site Catalog Skill site-catalog-skill
Learn how the Experience Modernization Agent’s site catalog skill performs automated analysis of an existing website to support Edge Delivery Services migration planning.
Overview overview
The site catalog skill discovers every page on the site, identifies the page templates and block variants in use, captures screenshots of each, and generates an interactive HTML report bundle you can browse in the Console preview tab or download and open locally.
The skill supports you and your migration of an existing project to Edge Delivery Services in the following ways:
- Starting a migration project — Run the skill before any work begins to understand the site’s scale including page count, templates, block variants, and locales. It establishes the baseline inventory every downstream decision depends on.
- Effort estimation and planning — Get quantified metrics to support proposals, sprint planning, and resourcing.
- Bulk import preparation — Use
template-catalog.jsonto identify which pages share the same layout and plan template-by-template bulk imports. - Stakeholder reporting — Share the interactive HTML report bundle with project managers, architects, and business stakeholders.
Invoking invoking
In the Experience Modernization Console, use natural language to ask the agent to catalog a site. The following are example prompts.
scope site https://www.example.comsite scope https://www.example.comanalyze https://www.example.comfind templates on https://www.example.comdiscover templates on https://www.example.comcatalog site https://www.example.comhow many page types are there on https://www.example.comwhat are the layouts on https://www.example.comanalyze site structure of https://www.example.com
You will notice that the skill’s workflow has four phases that run in sequence:
- Analyzing
- Templating
- Tuning
- Block cataloging
You can replay any phase and the agent clears that phase’s outputs and all downstream outputs and then resumes from that point forward. Here are some example prompts of replaying phases.
Repeat analyzing/Redo page analysis/Rerun analyze pagesRepeat templating/Redo the template discovery step/Restart the templating stepRepeat tuning/Rerun tune templates/Redo template tuningRepeat block cataloging/Restart catalog block variants
When replaying a phase, prior phases are preserved.
Output output
When the skill completes its cataloging of the site, you receive three different types of output.
-
A completion summary in chat including totals (pages, templates, block variants with EDS-mapped vs. custom breakdown), locale breakdown, coverage percentage, and overall report status (complete / incomplete / failed)
-
An interactive HTML report bundle as your primary deliverable, saved to
catalog/template-catalog-report-bundle.zip- The bundle contains
template-catalog-report.htmlplus all referenced screenshots and assets. - You can download the bundle and view it locally or share it.
- Or you can ask the agent to
Move template-catalog-report-bundle.zip to the /content folder to render it in the preview tab. Update all references as needed.to view the report in the console.
- The bundle contains
-
Structured JSON artifacts in
catalog/for downstream skills and programmatic use includingsummary.json,template-catalog.json,block-catalog.json,urls-all.json,urls-grouped.json,urls-checklist.json,.pages/,.blocks/
Catalog Folder Contents contents
Structured JSON artifacts are stored in catalog/ by the skill.
template-catalog-report-bundle.zipsummary.jsontemplate-catalog.jsonblock-catalog.jsonurls-all.jsonurls-grouped.jsonurls-sample.jsonurls-checklist.jsoncatalog.log.pages/<page-slug>/page-catalog.json.pages/<page-slug>/full-page.jpg.pages/<page-slug>/blocks/<block-name>.jpg.pages/_global/header.json + header.jpg.pages/_global/footer.json + footer.jpg.blocks/<variantId>/metadata.json.blocks/<variantId>/screenshots/<name>.jpgReport Statuses status
The status field in summary.json can be:
completeincompletefailedSampling for Large Sites sampling
By default, the skill limits deep page analysis to 1000 URLs. For sites with up to and including 1000 URLs, every page is analyzed.
For sites with more than 1000 URLs, the agent pauses and asks how to proceed:
- Increase the sampling cap (up to a maximum 4000 URLs)
- Analyze a specific group only (e.g. only
/products/*or/blog/*) - Analyze all URLs and run the full site with no sampling
URL discovery always covers the full site regardless of the sample limit. Only the deep per-page analysis phase is limited.
To override and analyze every page, tell the agent:
analyze all URLsanalyze everythinganalyze every pagerun the full site
Bulk Import Workflow bulk-import
The site catalog skill is part of the recommended approach for migrating a full site.
- Run the site catalog skill to get the full template catalog and block catalog.
- Open the HTML report bundle to visually review the templates the agent identified.
- For each template, manually import the representative pages (listed in
template-catalog.json) and refine the import until output is correct. - Bulk-import the remaining pages for that template using the URL list from
template-catalog.json. - Repeat for each template until the full site is migrated.
Limitations limitations
The site catalog skill has the following limitations.
- Public sites only — The target must be publicly accessible (no authentication, VPN, or firewall).
- Dynamic content is not supported — Content requiring user interaction to appear in the DOM may not be captured.
- Default 1000 URL limit - The deep analysis phase is by default limited to 1000 URLs, which can be overridden to up to a maximum of 4000 URLs.