Sovereign Credentialing Platform
Verifiable skill credentials for the uncredentialed.
🥈 VibeJam 2026 — 2nd Place · Choose Your Own Stack
-34d399?style=flat-square){kind=icon}
Self-taught developers, career-changers, and autodidacts face the same wall: employers don't trust skills without degrees. Multiple-choice certifications are gamed. GitHub repos are forked. Portfolio sites are templated.
Breakaway proves what you can actually do.
It is a zero-dependency, browser-native credentialing engine that generates tamper-proof, SHA-256 verified skill credentials through live cognitive challenges — code, debug, logic, and prompt engineering. No server. No database. No framework. Just you, the challenge, and cryptographic proof.
Built for people who can do the work but don't have the paper to prove it.
Real screenshots from the live application — no mockups.
The entry point. Dark obsidian background with VERITAS gold accents. One CTA: "Start Building Proof."
14 challenges across 5 domains and 3 difficulty tiers. Filter by domain, type, or difficulty. Each card shows available AI modes, XP reward, and time limit.
Cognitive radar chart (5-axis), skill trajectory, tier progress bars, and recent credentials. All data computed from completed challenges — no synthetic data.
Every completed challenge generates a SHA-256 verified credential. Stored locally. Each credential carries an integrity label (VERIFIED / CLEAN / OBSERVED / FLAGGED) based on behavioral signals.
Aggregated performance view: overall grade, domain breakdown, XP total, and skill map.
All metrics below are from a deterministic E2E validation run across all 14 challenges. Not estimated — measured.
| Metric | Value |
|---|---|
| Total Challenges | 14 |
| Challenge Types | 7 (code, quiz, validate, prompt, stroop, speedparse, aivalidate) |
| Domains Covered | 4 (Coding, Logic, AI Fluency, Debug) |
| Difficulty Tiers | 3 (Probe → Core → Edge) |
| Global Functions | 30 verified |
| Source Lines | 2,775 (373 HTML · 1,098 JS · 1,304 CSS) |
| Total Bundle | 140.7 KB (HTML + JS + CSS, unminified) |
| Benchmark Time | <1ms (full 12-challenge scoring pipeline) |
| Test | Result |
|---|---|
| Grade Boundaries (S/A/B/C/D/F) | ✅ PASS — 6 thresholds verified |
| Time Bonus Tiers (<30%/50%/70%/100%) | ✅ PASS — 4/4 correct |
| Tier XP Multipliers (0.8x/1.0x/1.4x) | ✅ PASS — 12/12 correct |
| SHA-256 Credential Hashing | ✅ PASS — 64-char hex, all unique |
| Radar Chart Data Parity | ✅ PASS — 5/5 axes match |
| Streak Calculation (1-hour window) | ✅ PASS — deterministic |
| localStorage Persistence | ✅ PASS — round-trip verified |
| Axis | Measures | Score |
|---|---|---|
| 🧮 Logic | code + quiz challenges | 79% |
| 👁 Attention | stroop cognitive interference | 80% |
| ⚡ Speed | speedparse code recall | 66% |
| 🔍 Validation | validate + aivalidate bug hunting | 80% |
| ✏️ Prompting | prompt engineering challenges | 75% |
| Grade | Count | Score Range |
|---|---|---|
| S (Sovereign) | 3 | 95–100% |
| A (Elite) | 3 | 85–94% |
| B (Proficient) | 3 | 70–84% |
| C (Developing) | 2 | 50–69% |
| D (Novice) | 1 | 25–49% |
| Tier | Challenges | Multiplier | Total XP |
|---|---|---|---|
| Probe | 4 | 0.8x | 510 |
| Core | 4 | 1.0x | 810 |
| Edge | 4 | 1.4x | 1,593 |
| Total | 12 | — | 2,913 |
| ID | Title | Type | Domain | Base XP | Time |
|---|---|---|---|---|---|
| c1 | Frequency Hunter | code |
Coding | 100 | 5:00 |
| c2 | Sequence Breaker | quiz |
Logic | 80 | 1:30 |
| c5 | The Prompt Architect | prompt |
AI Fluency | 180 | 5:00 |
| c8 | Snapshot | speedparse |
AI Fluency | 140 | 1:30 |
| ID | Title | Type | Domain | Base XP | Time |
|---|---|---|---|---|---|
| c3 | Mirror Mirror | code |
Coding | 150 | 4:00 |
| c4 | Ghost in the Code | validate |
AI Fluency | 200 | 7:00 |
| c6 | Test Saboteur | validate |
Debug | 175 | 6:00 |
| c7 | Signal & Noise | stroop |
Debug | 160 | 2:00 |
| ID | Title | Type | Domain | Base XP | Time |
|---|---|---|---|---|---|
| c9 | Hallucination Hunter | aivalidate |
AI Fluency | 250 | 8:00 |
| c10 | Refactor Reaper | code |
Coding | 280 | 8:00 |
| c11 | Temporal Debugger | quiz |
Logic | 220 | 2:00 |
| c12 | Architecture Critic | aivalidate |
AI Fluency | 320 | 10:00 |
breakaway/
├── index.html # 373-line SPA — 8 views, hash routing
├── js/
│ └── app.js # 1,098-line monolithic engine
├── css/
│ └── main.css # 1,304-line design system
└── assets/
└── preview*.png # UI screenshots
- HTML5 — Semantic SPA with 8 hash-routed views
- Vanilla JavaScript — Zero framework, zero build step
- CSS3 — Full design system with Inter + JetBrains Mono
- Web Crypto API — SHA-256 credential hashing via
crypto.subtle - localStorage — Client-side state persistence
#/ → Landing · #/dashboard → Dashboard · #/challenges → Gallery · #/challenge/:id → Active Challenge · #/results → Completion · #/credentials → Credential Vault · #/profile → User Profile · #/verify/:hash → Public Verification
No npm. No node_modules. No build pipeline. No framework. No CDN. Opens in any browser, works offline, ships as-is.
S (Sovereign) ≥ 95% → Gold tier
A (Elite) ≥ 85% → Green tier
B (Proficient) ≥ 70% → Cyan tier
C (Developing) ≥ 50% → Amber tier
D (Novice) ≥ 25% → Orange tier
F (Incomplete) < 25% → Red tier
< 30% of time used → +50 XP
< 50% of time used → +30 XP
< 70% of time used → +15 XP
≥ 70% of time used → +0 XP
Probe → 0.8x base XP (warm-up)
Core → 1.0x base XP (standard)
Edge → 1.4x base XP (elite)
Every completed challenge generates a verifiable credential:
hash = SHA-256(challengeId + code + score + time + aiMode + pasteCount + tabSwitches + timestamp)
Credentials include:
- Challenge ID, title, domain
- Raw score + grade
- Time bonus earned
- AI mode used (solo / augmented / orchestrator)
- Behavioral integrity label (VERIFIED / CLEAN / OBSERVED / FLAGGED)
- Paste count + tab-switch count (for solo mode)
- ISO timestamp
- 64-character hex hash
Credentials are stored locally and can be shared via the verification URL format: #/verify/:hash
Your credential reflects how you worked — not just what you produced.
When you select PURE HUMAN mode, Breakaway measures session continuity to give context to your credential. This is transparent, not judgmental — every signal is shown to you on the results screen.
| Signal | Method | Scope |
|---|---|---|
| Paste Events | onpaste listener on input fields |
Code editors, bug inputs, textareas |
| Tab Switches | visibilitychange API |
Entire challenge session |
| Typing Cadence | Keystroke interval analysis | Code challenges (solo mode only) |
| Label | Meaning | Color |
|---|---|---|
| FOCUSED | Zero interruptions — full session continuity | #34d399 (green) |
| STEADY | 1–2 minor signals — normal workflow | #06b6d4 (cyan) |
| SPLIT | 3–5 signals — attention was divided | #f59e0b (amber) |
| ASSISTED | 6+ signals — high external input detected | #ef4444 (red) |
- Informational, not gatekeeping — Every label is earned. An "ASSISTED" credential still proves you completed the challenge.
- Fully transparent — Exact paste counts and tab-switch counts are displayed on your results screen. Nothing is hidden.
- Deterministic —
computeIntegrity(pasteCount, tabSwitchCount, aiMode)is a pure function. Same inputs → same label. - Scope-limited — Only active in
solomode. AI-assisted modes (augmented,native) bypass tracking entirely. - Tamper-evident — Behavioral signal counts are inputs to the SHA-256 credential hash, so the label cannot be altered after the fact.
Note: The prior VERITAS seal (commit
e96f43d) has been superseded by a post-audit security patch.
A new full 10-gate pipeline run is required to re-seal the current commit.
Latest Audit: 2026-05-05 — Kimi Code CLI (VERITAS Ω Ecosystem)
Audited Commit: 8d13d0cc6953faaeb6487c94099a6fe41ed79a4f
Audit Seal: e34e55a82374f47056c3f81f543872dcecac45e58b8240920ecb937a410db619
Report: .veritas/audit-report-2026-05-05.md
| Issue | Fix |
|---|---|
| Stale VERITAS seal | Updated README with current commit + audit reference |
| Weak password hashing | Replaced SHA-256 with PBKDF2-HMAC-SHA-256 (100,000 iterations) + random 16-byte salt |
XSS via innerHTML |
Added escapeHtml() sanitizer for all user profile data |
| Duplicate toast logic | Removed duplicate rank-up/streak toast block in showResults() |
| Gate | Verdict |
|---|---|
| INTAKE | ✅ PASS |
| TYPE | ✅ PASS |
| DEPENDENCY | ✅ PASS |
| EVIDENCE | ✅ PASS |
| MATH | ✅ PASS |
| COST | ✅ PASS |
| INCENTIVE | ✅ PASS |
| SECURITY | 🟡 PENDING RE-SEAL |
| ADVERSARY | 🟡 PENDING RE-SEAL |
| TRACE/SEAL | 🟡 PENDING RE-SEAL |
# Clone
git clone https://github.com/VrtxOmega/breakaway.git
# Open
open breakaway/index.html
# or just double-click index.htmlNo install. No build. No server required.
| Platform | URL |
|---|---|
| Live Demo | vrtxomega.github.io/breakaway |
| GitHub | VrtxOmega/breakaway |
| Codeberg | VeritasOmega/breakaway |
breakaway/
├── index.html # 430-line SPA — 8 views, hash routing
├── js/
│ └── app.js # 1,200-line engine (being modularized)
├── css/
│ └── main.css # 1,400-line design system
├── assets/
│ └── preview*.png # UI screenshots
└── .veritas/
└── audit-report-2026-05-05.md
Stack: HTML5 · Vanilla JS · CSS3 · Web Crypto API · localStorage
Constraint: Zero dependencies. Zero build step. Works offline.
Breakaway is part of the VERITAS Omega Universe. See VERITAS.md for the full ecosystem map, gate pipeline spec, and sister projects.
MIT