Skip to main content

How it works

Real environments.
Real proof.

Candidates build a verified skill wall through real scenarios and live technical challenges. Recruiters search ranked, anonymous walls and pay credits only when they commit to making contact. The whole platform runs on four pillars.

Pillar 1

The Wall

Anonymous by default. Every verified skill is a brick; every brick carries the full proof trail. Walls grow across your whole career and you never lose them.

Pillar 2

Real-environment assessment

Kubernetes challenges run in real k3s namespaces; Linux challenges in real containers. A real-time agent observes, reasons, and asks. Process intelligence is measured, not just outcomes.

Pillar 3

Paid unlock

Recruiters pay a credit to reveal a candidate's identity. Messaging stays in-platform; contact never leaks to email threads we cannot moderate.

Pillar 4

JD-driven ranking

Every job description surfaces matched candidates plus the gaps between them and the role. Closing a gap moves a candidate up the live ranking for that JD.

Side by side

Two flows, one platform.

Candidate flow

Prove, earn, grow.

  1. Step 1

    Sign up with email and a non-VoIP phone number. Your wall is created immediately; you stay anonymous to recruiters until you decide otherwise.

  2. Step 2

    Pick a topic and take the assessment. Five stages, scored by weight: scenario, probe, live task, command probe, process intelligence throughout.

  3. Step 3

    Earn a brick at one of five tiers with the full proof trail attached. Re-test or take harder challenges later to upgrade the tier.

  4. Step 4

    Paste a job description and see the gap; close the gap to climb the ranking for that role. Decide whether to accept unlock requests one by one.

Candidate overview →

Recruiter flow

Search, unlock, converse.

  1. Step 1

    Apply with your work email. Early applications are manually reviewed; approval is by invitation during private beta.

  2. Step 2

    Post a JD or run a direct search. Filter by skill, tier, and score; scan anonymous walls and the proof trail behind every brick.

  3. Step 3

    Spend a credit to unlock a candidate's identity permanently. Message them in-platform; they choose whether to reply.

  4. Step 4

    If they do not respond within 14 days, 100% of the credit returns to your balance automatically. Credits are valid for 12 months.

Recruiter overview →

Assessment

Five stages, weighted toward what is hard to fake.

Your final score is weighted toward the live task, because that is the stage that actually demonstrates skill. The calibration and probe stages catch overclaiming; the command probe catches lucky keystrokes.

  1. Stage 1

    Calibration

    Sets difficulty

    A short calibration sets the right difficulty so the real assessment is neither too hard nor too soft. You get a preview of what the next stages will look like.

  2. Stage 2

    Scenario question

    15%

    A real-world situation. You describe how you would approach it in your own words. No multiple choice, no fill-in-the-blank, no stock answers.

  3. Stage 3

    Follow-up probe

    10%

    An AI examiner asks follow-up questions tied to your specific answer. Hand-waving gets caught; real thinking is rewarded. No canned interviewer template.

  4. Stage 4

    Live task in a real environment

    65%

    A real k3s namespace or Linux container with something broken. You use your own commands in a real terminal. A real-time agent observes your process and asks targeted questions. It never takes actions for you.

  5. Stage 5

    Command probe

    10%

    The examiner asks why you ran the commands you ran. Lucky keystrokes surface here; diagnostic sequence is rewarded. Recovery from mistakes counts.

Process-intelligence signals are measured throughout the live task rather than as a separate stage, so you do not context-switch out of debugging to take a quiz.

Tiers

Five tiers, one continuous scale.

Every brick lands in one of five tiers. The same skill can appear at different tiers across your career as you re-test or take on harder challenges.

  • Raw

    Score 0+

    Skill claimed but not yet tested.

  • Solid

    Score 50+

    Baseline proof — scenario question passed.

  • Forged

    Score 70+

    Deep proof — live task completed in a real environment.

  • Polished

    Score 85+

    Expert level — high score with strong process intelligence.

  • Diamond

    Score 95+

    Elite — top 5% of platform, consistent across retests.

Privacy by design

Anonymous until you decide otherwise.

  • Public wall URLs use a random reference, never a real name. Caches, logs, and analytics are PII-scrubbed at ingestion.
  • Identity unlocks are event-logged. You can see which recruiter unlocked you and when; they cannot message you off-platform.
  • Self-service data export and erasure; 30-day response window for UK and EU GDPR requests. See the privacy policy for the full picture.

Questions

How it works, in detail.

How long does a full assessment take?
Budget roughly 30 minutes for the live task plus the scenario and probe stages. The hard wall-clock cap is 1x the intended duration, so you always have headroom. Short disconnects pause the timer.
What happens if I lose connection?
Up to 30 minutes of disconnect per session is handled gracefully. Your shell survives, your timer pauses, and your progress is preserved. More than 2 disconnects inside a single session ends the run and we record the partial outcome.
Why use real environments instead of simulations?
Because simulators reward memorised commands; real environments reward debugging. You run kubectl against a live cluster, inspect real logs, hit real rate limits. The agent observes the sequence you take, not just whether the screen ends up green.
What does "process intelligence" score?
Diagnostic sequence, command efficiency, hypothesis formation, verification discipline, and recovery from mistakes. The signals a senior engineer notices when watching someone work - measured continuously across the task, not as a separate quiz.
How does GitHub fit in?
Display-only context. The examiner may cross-reference your public repos during probes, but GitHub activity never feeds your tier or score. We have seen too many "green-square" CVs to treat commit streaks as evidence of skill.
When do recruiters see my name?
Only after they spend a credit to unlock you. Before that, everything about your wall is anonymous: search results, public wall URLs, cache keys, logs, analytics. Messaging stays in-platform; you control every reply.

Try an assessment

Free forever. No credit card. Prove one skill and see the shape of it.

Create your wall

Hiring engineers?

See credit packs, subscriptions, and how the ghost-refund guarantee works.

See pricing