Reference Level: Advanced
Use and reuse this content
The portal is built to be reused. Pick the approach that fits how your organisation shares knowledge, from a simple link on your intranet to your own branded copy, an AI assistant or an integration through the Model Context Protocol (MCP). Content is published under the Open Government Licence and code under the MIT licence.
In brief
- Ten approaches, from a link on your intranet to your own branded copy.
- AI options: an assistant that cites the portal, and a reference MCP server.
- Each approach says what it takes and what to watch out for.
Approaches
- Low effort
U-01 · Link and embed
Link from your intranet
Stable addresses for every page, tab, step and capability.
How
- Link to a page, or to a section with its anchor, for example governance.html#method.
- Share links with the Share button on any page.
- Point staff to the role routes and the testing page.
Choose this when: You want staff to use the guidance as it is, always current.
Watch out for
- Links go to a prototype that is not official guidance; say so on your intranet.
- Low effort
U-02 · Link and embed
Embed pages in your intranet or Teams
Show portal pages inside SharePoint, an intranet page or a Microsoft Teams tab.
How
- Add a web part or a Teams website tab pointing at a page address.
- Choose pages that work well on their own, such as a playbook or the self-assessment.
- Keep the prototype banner visible.
Choose this when: You want the content inside the tools people already use.
Watch out for
- Progress and display choices are kept per browser, so they do not follow people between devices.
- Low effort
U-03 · Data and APIs
Read-only data API
Every dataset is a stable file you can fetch from code: JSON, CSV, Turtle and YAML, with cross-site access allowed.
How
- Start from the catalogue at data/catalogue.jsonld.
- Fetch data/playbooks.json for the full content, or a single CSV.
- Pin to a release by recording the version in the catalogue.
Choose this when: You want to build your own views, dashboards or tools on the content.
Watch out for
- Files change with each release; check the version and the What is new page before relying on a field.
- Low effort
U-04 · Data and APIs
Harvest into a data catalogue
A DCAT 3 catalogue other catalogues can harvest, such as CKAN or your own metadata platform.
How
- Point your harvester at data/catalogue.jsonld or data/catalogue.ttl.
- Map dataset identifiers to your own scheme.
Choose this when: You run a catalogue and want the playbook datasets listed alongside your own.
Watch out for
- Treat harvested entries as reference material, not as your own critical assets.
- Medium effort
U-05 · Enterprise content
Load into enterprise content and collaboration tools
Bring templates, checklists and actions into SharePoint lists, Confluence, Planner or your tracker.
How
- Import a CSV, such as change-checklist.csv or templates.csv, into a list or board.
- Link each item back to its page for the full guidance.
- Agree an owner to refresh it after each release.
Choose this when: You want teams to work from tasks and templates in their own tools.
Watch out for
- Copies go out of date; note the version you imported.
- Medium effort
U-06 · Adapt and brand
White-label your own copy
Run your own branded copy with your organisation's name, colours, tone and extra content.
How
- Fork the repository on GitHub.
- Change the title, contact form and site address in the content file, and colours in the style sheet.
- Edit or add content in the content files; the build checks every reference.
- Publish with the included workflow on your own GitHub Pages or web server.
Choose this when: You need the guidance in your own house style, or with local procedures added.
Watch out for
- Keep the attribution statement and the licences.
- Your copy no longer gets updates automatically; merge new releases deliberately.
- Medium effort
U-07 · Adapt and brand
Adapt for your audiences and language
Tailor wording, examples and depth for particular audiences, or translate, for example into Welsh.
How
- Copy the content files and adjust text, keeping identifiers so links still work.
- Use the plain English and style checks in the build to keep quality up.
Choose this when: Your audiences need different language, tone or examples.
Watch out for
- Translations need a named owner to keep them in step with releases.
- Low effort
U-08 · AI and integrations
Ask an AI assistant about the portal
Use an AI assistant or chatbot to answer questions from the portal, alongside the built-in search.
How
- Give the assistant the address of llms.txt, a short guide written for AI tools, or llms-full.txt with the full text.
- Or add the portal website, or data/playbooks.json, as a knowledge source in a tool such as Microsoft Copilot Studio.
- Ask it to cite the page each answer comes from.
Choose this when: You want people to ask questions in their own words.
Watch out for
- AI answers can be wrong; check them against the page cited, and against GOV.UK, which is the authority.
- Do not paste personal or sensitive data into public AI tools.
- Medium effort
U-09 · AI and integrations
Connect through the Model Context Protocol (MCP)
A small reference MCP server in the repository lets AI assistants search the portal and fetch capabilities, requirements, practices and sector rules as structured data, with links to their pages.
How
- Run integrations/mcp/server.py with Python 3; it needs no extra packages.
- Add it as an MCP server in an MCP-compatible assistant.
- Point it at the live site or at a local copy of data/playbooks.json.
Choose this when: You want an assistant or agent to use the portal's content as a tool, with sources.
Watch out for
- It is a reference implementation for testing, not a supported service.
- Host it inside your own environment if you add internal content.
- Medium effort
U-10 · AI and integrations
Connectors, add-ons and plugins
Use the open data and the MCP server as the basis for a connector, add-on or plugin in your organisation's platforms.
How
- Use the catalogue and JSON files as the data source.
- Wrap the MCP server or the files in your platform's connector or plugin format.
- Register it through your organisation's approval process for integrations.
Choose this when: You want the content inside a business application or an AI assistant your organisation already uses.
Watch out for
- Follow your organisation's rules for approving third-party integrations.
Embed a requirement or capability
Show one DAM Policy requirement or one capability on your intranet or in a document that accepts web content. Each has its own small page, with a link back here. Copy this code and change the identifier, for example DAM-R01 or CR-01:
<iframe src="[this portal's address]/embed/DAM-R01.html" title="DAM Policy requirement DAM-R01" width="100%" height="220" loading="lazy"></iframe>
Preview: DAM-R01 · CR-01. Embedded pages do not change when your copy is cached; they always show the latest release.
Offline copy
If your network blocks this site, download the whole portal as a zip file, unzip it, and open index.html in your browser. Search, filters and progress work offline. The copy is for this release only: check What's new for updates.
Addresses for tools and AI
| Address | What it is |
|---|---|
data/catalogue.jsonld | DCAT catalogue of every dataset (JSON-LD) |
data/catalogue.ttl | The same catalogue in Turtle |
data/playbooks.json | All content in one file |
llms.txt | Short guide for AI tools |
llms-full.txt | Full text for AI tools |
sitemap.xml | Every page address |
feed.xml | Releases as an Atom feed |
data/ontology.ttl | Ontology of the portal (OWL, Turtle) |
data/lineage.ttl | Lineage from sources to requirements, pages and checks (PROV-O, Turtle) |
AI assistants and search
The built-in search finds pages, steps and terms by their words. An AI assistant can answer questions in plain language, combining several pages. This portal does not run its own chatbot: there is no server, and nothing you type is sent anywhere. Instead it publishes its content in forms that assistants your organisation already approves can use, with the page each answer comes from.
- llms.txt is a short, structured guide to the portal for AI tools, following a common convention; llms-full.txt holds the full text.
- Knowledge sources: add the site address or
data/playbooks.jsonto an assistant builder, such as Microsoft Copilot Studio. - Always check an answer against the page it cites, and against GOV.UK, which remains the authority.
Model Context Protocol (MCP) server
A reference server in the repository, integrations/mcp/server.py, lets an MCP-compatible assistant use the portal as a set of tools. It uses only the Python standard library, reads the published content, and returns links to the pages it draws on. It is for testing and adaptation, not a supported service.
| Tool | What it does |
|---|---|
search_portal | Search steps, requirements, capabilities, practices, sector supplements and glossary terms. |
get_capability | Get one capability from the capability model, with its signposts. |
get_requirement | Get one DAM Policy requirement with its source and strength. |
get_practice | Get one delivery and change practice with its actions. |
list_sector_rules | List the quality rules for a sector supplement. |
Download the full content file, or set DAM_PLAYBOOKS_URL to its full address to read the live copy. Example configuration for an assistant that reads MCP servers from a JSON file:
{
"mcpServers": {
"dam-playbooks": {
"command": "python3",
"args": ["integrations/mcp/server.py"],
"env": {"DAM_PLAYBOOKS_FILE": "/path/to/playbooks.json"}
}
}
}
Whatever route you choose, keep the prototype status, the attribution statement and the licences with the content, and agree who in your organisation keeps your copy up to date.
Was this page useful? Tell us Tell us privately, without GitHub