MCP-powered content intelligence

CSS Knowledge Insights

Filter live MCP articles by source or repo, then review quality, freshness, state, links, and action paths.

0filtered articles

0matching repos

0page-view signal

0%avg quality

0broken or unchecked links

Overall content health

78%

Live computed from MCP source results

Needs action

Calculated from stale, critical, and metadata gaps

Freshness signal

84%

Uses Last Updated where exposed by source content

MCP connector state

4/4

Checking source MCP endpoints

Source system health

MCP normalizes signals from each source into freshness, trust, usage, and actionability scores.

Live model

CSS journey heatmap

A fast scan of where CSS users are most likely to encounter stale or unclear content.

Live MCP tracked metrics

Live values derived from connected MCP source results, with gaps called out where page-view or cost telemetry is not exposed by MCP.

MCP live

All articles retrieved from MCP

Primary review pane with clickable articles, selected filter context, and sortable quality signals.

0 articles

Source and repo health

Repo/source rollup from retrieved MCP articles using quality score, freshness, state, and link health.

0 repos

Highest quality retrieved articles

Best-scoring live MCP articles by metadata completeness, freshness, actionability, state, and link status.

High-impact content needing attention

Identified from MCP source results using relevance, missing freshness metadata, actionability terms, known-issue/case-support language, and source health score.

How to read the action labels

Plain-language meaning for the actions shown in the high-impact queue.

Add freshness signal

Add or expose Last Updated, next review date, owner, and freshness SLA so users know if the page is current.

Validate accuracy

Ask the content owner or SME to confirm that guidance matches current product behavior and support process.

Improve actionability

Rewrite the content so CSS users have a clear next step, resolution path, escalation route, or handoff.

MCP connector flow

Connector role and transformation stages.

Ingest source systems

LMC, SMC, Evergreen, CSS Wiki

Normalize metadata

Owner, persona, journey, last reviewed, duplicate cluster

Score health

Freshness, accuracy, usefulness, findability, actionability

Watch

Route action

Content owner backlog, review queue, retirement candidates

Recommended actions

What the CSS content team should do next.

Where users take action

Route each health finding into the right Content Hub, Rubik, Content Mentor, PR, or ADO workflow.

Health model

Suggested scoring dimensions for the first MCP-backed release.

Freshness

Last reviewed, source update age, policy drift.

25%

Accuracy

Contradictions, SME confidence, known issue links.

25%

Usage signal

Views, failed searches, case deflection, escalations.

20%

Findability

Search rank, metadata quality, duplicate overlap.

15%

Actionability

Clear next step, owner, workflow, resolution path.

15%

Run with node content-health-dashboard-server.js to use live MCP data. Evergreen and CSS Wiki may require environment variables for authorization or endpoint configuration.