Skillbanc AI : Examiner turns the whole certification blueprint into a graph of cues and answers — and has you reconstruct it from memory, link by link. Every wrong link is corrected on the spot. It's active recall, not passive reading, so what you study actually stays.
Android app coming soon to Google Play · First certification: Claude Certified Architect — Foundations (CCAR-F)
The exam never asks you to recite a paragraph. It asks you to make a decision: given this situation, what's the right move? So that's what you practice — one compressed link at a time.
You see the left node. You build the link. Miss it, and the right answer appears in place — with the one-line reason why.
Retrieving an answer from memory is what builds durable memory — the testing effect. Highlighting a PDF feels productive but fades fast. Every drill here is a retrieval.
Constructing the connection yourself encodes it far deeper than being shown it — the generation effect. You're not reading the graph; you're rebuilding it.
Each fact is tied to related facts, contrasts, and traps. More connections mean more paths back to the answer under exam pressure — elaborative encoding.
A wrong link isn't marked and forgotten. The correct edge and its reason surface immediately, so the misconception never has time to set.
Links you miss come back sooner; links you've proven get periodic spot-checks. Your minutes flow to the edges that are actually shaky.
Time-on-domain is tracked silently as you work — so for the first time you can answer "how many hours does it really take to get ready?" with data, not a guess.
Most certification prep is built around consumption — long PDFs, hours of video, and a question bank you eventually memorize. The Examiner is built around retrieval.
Every link you build and every question you answer updates a live per-domain mastery model. The Examiner always knows exactly where you stand.
Rebuild the exam's decision graph from memory — situations, contrasts, rules, and traps — with minimal typing and instant, in-place correction.
Each domain has a live mastery score that moves with every link and answer — forge sessions, check-ins, and mocks all feed one model.
The Examiner steers you toward weak points and away from what you've proven — with periodic spot-checks so strong domains don't silently decay.
Fresh scenario-style questions in the official exam's voice, generated from the public blueprint — blueprint-weighted, weak-domain-biased, explanation on every answer.
Every drill is timed and logged behind the scenes, per domain — evidence of exactly how long readiness took, and a benchmark against the average.
Certification isn't a feeling: 80%+ overall and 70%+ per domain, twice in a row, on full-length exams. The Examiner won't call you ready until you are.
No 120-minute slogs while you're learning. The Examiner assesses you continuously and spends your time only where it moves your score.
Rebuild each domain's cue→answer links from memory in the Concept Forge. Miss one and the right link is corrected on the spot.
Short 12-question sessions weighted to your weak domains. Strong areas still get probed — so proven strength stays proven.
When you're close, the Examiner writes a fresh, full-length, blueprint-weighted exam targeting your remaining gaps. Never the same questions twice.
Two consecutive passes with every domain above the floor and the AI Examiner certifies you. Then — and only then — book the real exam.
CCAR-F validates real-world judgment building with Claude Code, the Agent SDK, the Claude API, and MCP. The Examiner ships with the complete public blueprint, reshaped for recall:
Sign in, forge your first domain, and watch your mastery map fill in — in minutes. The Android app lands on Google Play soon.