Digital Twin Learner Model

Domain Heatmap

One cell per top-level group in the knowledge graph. Background tint = mastery percent. The purple bar inside each cell = average FSRS stability of cards in that group (longer = more durable). Hover for details.

Group Detail

Click a cell above to drill into a group. Recent activity = the latest mastery toggle (from FSRS history we can detect review dates only — mastery toggles aren't timestamped in nexus_mastery_v1, so "last touch" prefers the latest review for that group).

Spaced-Repetition Health

Composite 0-100 score across four sub-dimensions. Higher = stronger long-term retention discipline. The dial is a heuristic, not a clinical metric — interpret it as a steering signal, not a verdict.

Card Pool

Lifetime rating distribution

Forgetting-Curve Estimate (Ebbinghaus / FSRS-style)

For cards last reviewed N days ago, the Nexus FSRS scheduler targets ~90% retention by next due date. This panel buckets cards by days since last review and shows the share above the target retention line — a rough "would I still remember it now?" proxy.

Bucket label is the median age of cards in the bucket. Retention estimate uses the FSRS power-law approximation R(t) = (1 + t/(9·S))−1 with the card's own stability S; reported figure is the share of cards in the bucket whose estimated retention is still ≥0.90 (the FSRS default target).

Streak

Domain-Specific Recall (DSR) Recommendations

Algorithm

1. From the knowledge graph, find every unmastered concept whose direct prerequisites are all mastered (BFS frontier — same as Skill Mastery Map's "Next to Study" tab).
2. For each candidate, compute three signals:
   • Domain priority bonus — your three weakest domains (lowest mastery %, ≥3 nodes) get +5 each. Pulls study toward your blind spots instead of letting strongholds keep growing.
   • Downstream impact — BFS count of dependent concepts (capped at 200). Score = log₂(impact + 1) × 3. Concentrating on high-fan-out concepts compounds.
   • Prerequisite-of-soon-due — for each FSRS card in your pool that is due within 7 days and depends on this candidate concept, +2. Studying a prerequisite right before its dependent is reviewed reinforces the chain.
3. Rank by total score, return top 10. Each recommendation comes with an explanation of which signals fired.

Why this heuristic

Pure spaced-repetition (review whatever's due) optimizes for short-term retention but is blind to which gap matters more. Pure prerequisite-frontier traversal (Skill Mastery Map default) is gap-aware but doesn't account for which gaps are load-bearing for the rest of the graph or for your near-future review queue. DSR here combines all three signals in one rank — your weakest domain + structural impact + temporal urgency.

Honest limits

• The "domain priority" floor (3-node minimum) means a tiny custom group with one mastered concept won't dominate the bonus. Real curricula vary; tune to your needs.
• The "prerequisite-of-soon-due" signal needs a non-trivial FSRS pool to fire. With <30 cards scheduled, that signal will mostly be zero.
• "Downstream impact" assumes the graph topology reflects real dependency. Edges in graph.json are curated; gaps in curation under-credit some concepts.

Computed Insights

What this profile cannot see

• Knowledge gained outside Nexus (other courses, certifications, books, blogs).
• Skills demonstrated in real production work or incident response.
• Industry certifications already held (CISSP, OSCP, GIAC, AWS, Azure, etc.) unless their content overlaps a tracked concept and you also marked it mastered.
• Hands-on lab performance not recorded in mastery toggles.
• Conceptual depth that you have but never bothered to flag — un-marked concepts look identical to un-known concepts to this tool.
• Whether the FSRS card's underlying question matches the depth you actually understand. Rating Good on a recall card doesn't prove you can apply or design.

Privacy and data

localStorage stays local. No data leaves your browser. The only network call this page makes is a same-origin fetch of ../learning-graph/graph.json. There is no Plausible or other analytics on this page, no third-party fonts, no CDN scripts.

Read-only on sibling keys. This tool never writes to nexus_mastery_v1, nexus_fsrs_v1, nexus_fsrs_history_v1, or nexus_portfolio_v1. You can verify by opening DevTools → Application → Local Storage and watching them while you click around.

Self-track over time

The export below produces a JSON snapshot of your computed profile at the current moment. Save snapshots periodically and diff them yourself to see your trajectory — this tool deliberately doesn't store history because that would conflict with the read-only / minimal-storage stance.

Data sources detected