rene.fichtmueller/shieldx
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
20 Commits 3 Branches 0 Tags
Commit Graph

20 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Claude
cd31274738
feat(ai-act-kompass): i18n, national modules, task checklist, legal gate, new documents 
- Full i18n (DE/EN): UI, rules engine, content pack and generated
  documents; language switcher, browser-language default, architecture
  extensible to further EU languages via localized source texts
- National country modules (opt-in per company profile): DE (KI-MIG/
  BNetzA supervision, works council co-determination Sec. 87 BetrVG,
  GDPR/BDSG), AT (RTR AI service desk), IT (Law 132/2025: employer info
  duty, deepfake criminal liability, sector decrees), ES (AESIA,
  sandbox RD 817/2023), FR (CNIL guidance), NL (algorithm register);
  merged into obligation derivation, checklists and documents
- New EU obligations: corrective actions (Art. 20/21), authorised
  representative (Art. 22), EU database registration for public-body
  deployers (Art. 49(3)), post-market monitoring (Art. 72), serious
  incident reporting (Art. 73), GDPR DPIA (Art. 35 GDPR / 26(9))
- New documents: AI procurement requirements specification (MUST/SHOULD
  matrix for vendors) and EU declaration of conformity draft (Annex V)
- Cross-system checkable task list (due now / upcoming / done) with
  progress bar, one-tap toggling, per-system links
- Legal safety: first-run acknowledgement gate (working aid, no legal
  services), legal notice panel in settings, result-screen disclaimer,
  disclaimers on every document
- Responsive layout for phone/tablet (top-bar nav, scrollable tables,
  larger tap targets)
- 12 new tests (44 total): EN pack parity, national merge, checklist
  grouping, new document generators

Claude-Session: https://claude.ai/code/session_017YjohVNx1ZGNM4MX7tHpfv
 2026-07-04 09:18:07 +00:00
Claude
06574c5cfd
feat(ai-act-kompass): anti-AI print design and company personalization 
- Full redesign: paper tone, serif typography, hairlines, rectangular
  stamp badges, mono labels — technical print look instead of SaaS style,
  no gradients/shadows/emoji
- Company profile settings (name, address, contact person, email, phone,
  website, logo upload as local data URL) persisted in localStorage and
  included in JSON backup (schema v2, v1 imports still supported)
- Print/PDF output via letterhead print view (window.print): logo,
  company name and contact data in header, contact line in footer;
  applies to risk report, action plan, Annex IV skeleton and inventory
- Company block also embedded in generated Markdown headers
- Minimal auditable Markdown renderer (no external parser), HTML-escaped
- 8 new tests (32 total)

Claude-Session: https://claude.ai/code/session_017YjohVNx1ZGNM4MX7tHpfv
 2026-07-04 09:18:07 +00:00
Claude
fb2da0db00
feat: add AI-Act-Kompass — standalone EU AI Act compliance toolkit for SMEs 
Local-first React/Vite app (no cloud, no telemetry, localStorage only):
- Assessment wizard classifying AI systems per AI Act systematics
  (Art. 3 definition, Art. 5 prohibitions, Art. 6/Annex I+III high-risk
  incl. Art. 6(3) exception and profiling bar, Art. 50 transparency, GPAI)
- Role-specific obligation checklists (provider/deployer/importer/distributor)
  with legal basis, deadlines and implementation status
- Document generator: risk report, action plan, Annex IV skeleton (Markdown)
- AI inventory with export, deadline overview incl. Digital Omnibus shift
- Versioned regulatory content pack modeled as data for future update feed
- Engine fully unit-tested (24 tests), TS strict, static build

Claude-Session: https://claude.ai/code/session_017YjohVNx1ZGNM4MX7tHpfv
 2026-07-04 08:03:16 +00:00
renefichtmueller
874e4a1ead Security: scrub home IPs, internal domains, name/email + local paths (files + history)
All checks were successful
security-scan / secret-scan (push) Successful in 3s
Details
2026-06-23 23:20:59 -01:00
Rene Fichtmueller (CtX)
ecac7c865e CI: direct gitleaks full-scan (robust to force-push); IP check skips docs/tests 2026-06-23 22:57:11 -01:00
renefichtmueller
cbde0363a6 Security: scrub host/server/org identifiers + token from files and history; add scan allowlists for the project's own public names and test corpus 2026-06-23 22:51:34 -01:00
Rene Fichtmueller (CtX)
266d45733d Upgrade CI to gitleaks (entropy + curated rules, max heuristic) 2026-06-23 22:36:15 -01:00
Rene Fichtmueller (CtX)
709e47f1de Add CI secret-scan safeguard (blocks secrets/private-IPs/key-files on push+PR) 2026-06-23 20:56:27 -01:00
Rene Fichtmueller
27df0f09fb docs: update descriptions with typoglycemia, 547+ rules, 50+ languages 2026-04-07 11:57:58 +02:00
Rene Fichtmueller
338e1c5b92 feat: add typoglycemia detection to TokenizerNormalizer 
Detects scrambled-middle-letter attack words (OWASP LLM defense).
Pre-computed signature map for O(1) lookups — "ignroe" → "ignore",
"bypssa" → "bypass", "insrtuctinos" → "instructions".
40 attack keywords covered. Zero false positives on benchmark.
 2026-04-07 11:35:10 +02:00
Rene Fichtmueller
d12b366cdc docs: update changelog with multilingual expansion stats 2026-04-07 01:09:54 +02:00
Rene Fichtmueller
221c1d4868 feat: expand multilingual detection to 211 rules across 50+ languages 
- TPR improved from 70.8% to 91.9% (324 sample benchmark)
- Multilingual attack TPR: 96.6% (29 samples)
- Deep South Asian coverage: Bengali (9), Hindi (8), Urdu (6), Tamil (4),
  Telugu (3), Marathi (4), Gujarati (3), Kannada (2), Malayalam (2),
  Punjabi (2), Sinhala (2), Nepali (4), Pan-Indic transliterated (7)
- New languages: Persian, Hebrew, Kurdish, Indonesian, Filipino, Burmese,
  Khmer, Lao, Finnish, Czech, Slovak, Romanian, Hungarian, Greek, Bulgarian,
  Croatian, Serbian, Georgian, Armenian, Azerbaijani, Swahili, Amharic,
  Afrikaans, Mongolian, and 20+ more
- Universal patterns: rapid script switching, global DAN mode, cross-script
  password extraction, no-filter patterns
- README updated with new benchmark results and language coverage tables
 2026-04-07 01:08:09 +02:00
Rene Fichtmueller
7b26bdec56 docs: fix clone URL for public repo 2026-04-07 00:36:46 +02:00
Rene Fichtmueller
2a39ea4362 docs: comprehensive v0.5.0 README with full feature documentation 
- Architecture diagram updated with all new modules (ensemble, ATLAS, evolution, immune memory)
- Benchmark results section (70.8% TPR, 0.0% FPR)
- Defense modules overview table with line counts
- 369+ detection rules across 12 categories documented
- Bio-immune self-evolution (6 mechanisms) fully explained
- Preprocessing pipeline: CipherDecoder, TokenizerNormalizer, Unicode
- MITRE ATLAS mapping (90 techniques, 8 tactics) with API examples
- MCP Guard with MELON, tool chain, resource governor details
- Decomposition attack detection documentation
- Supply chain integrity section
- Multilingual detection (20+ languages) with examples
- RAG Shield documentation
- Output validation and OutputPayloadGuard docs
- Compliance section (MITRE ATLAS, OWASP LLM Top 10, EU AI Act)
- Full project structure tree
- Updated feature comparison table (30 features vs competitors)
- Updated performance targets with new modules
- Bio-immune API examples (evolution, adversarial training, calibration)
- 1265 lines from 604 — over 2x content increase
 2026-04-07 00:36:20 +02:00
Rene Fichtmueller
371d3829e8 feat: ShieldX v0.5.0 — full defense evolution + pentest hardening 
4-phase defense evolution (Bio-Immune, Adversarial, Ensemble, ATLAS)
with ~200 new detection rules across 20 languages.

TPR 32.9% → 70.8%, FPR 12.2% → 0.0%

New modules: DefenseEnsemble, AtlasTechniqueMapper, EvolutionEngine,
ImmuneMemory, FeverResponse, MELONGuard, AdversarialTrainer,
DecompositionDetector, IndirectInjectionDetector, OutputPayloadGuard,
ToolCallSafetyGuard, AuthContextGuard, ResourceExhaustionDetector,
TokenizerDeobfuscation, Binary/Hex decoder, OverDefenseCalibrator
 2026-04-07 00:27:12 +02:00
Rene Fichtmueller
844c8d90d0 fix: remove local file paths from research reference 2026-04-04 23:07:35 +02:00
Rene Fichtmueller
9881dd3bb5 feat(security): v0.4.0 — three research-driven detection gaps closed 
Implements hardening based on sarendis56/Jailbreak_Detection_RCS
(arXiv:2512.12069) and the Awesome-LVLM-Attack/Safety survey series.

L0 — CipherDecoder: FlipAttack, ROT13, Caesar (all 25 shifts), Morse,
Leet speak, Pig Latin, ASCII art detection with suspicion scoring.

L2 — SemanticContrastiveScanner: RCS-style harmful/benign bucket
comparison via EmbeddingStore, 20 canonical jailbreak seeds, BoW
embedding fallback for offline use.

L6 — ConversationTracker: Crescendo (+0.35), Foot-in-the-Door (+0.40),
Jigsaw Puzzle (+0.45) multi-turn escalation patterns added.

292/294 tests passing (2 pre-existing ATLASMapper failures unrelated).
 2026-04-04 23:04:42 +02:00
Rene Fichtmueller
bacdd344a7 feat(rules): mcp-007..010 — Claude Code source map leak countermeasures 
Rules based on 2026-03-31 npm source map disclosure:
- mcp-007: Coordinator Mode / KAIROS / ULTRAPLAN invocation attempts
- mcp-008: Multi-agent spawn manipulation via known spawning mechanism
- mcp-009: Persistent memory file targeting (CLAUDE.md / .claude/ injection)
- mcp-010: Tool enumeration probe (reconnaissance of available tools)

Source: github.com/Kuberwastaken/claude-code, @anthropic-ai/claude-code
MITRE ATLAS: AML.T0062, AML.T0051, AML.TA0015 (C2)
 2026-03-31 16:54:29 +02:00
Rene Fichtmueller
4967007ab3 feat(scripts): daily arXiv + HackerNews security monitor 
Autonomous monitoring script for Erik VPS:
- Fetches arXiv cs.CR + cs.AI RSS feeds daily
- Fetches HackerNews top stories + keyword RSS feeds
- Classifies relevance via Claude Haiku API (HIGH/MEDIUM/LOW/SKIP)
- HIGH findings: generates TypeScript detection rules via Claude
- Appends rules to src/detection/AutoGeneratedRules.ts
- Runs tsc --noEmit before committing (zero errors required)
- Commits + pushes to Gitea on success
- JSON report saved to /opt/scripts/logs/shieldx-report-YYYY-MM-DD.json
- Cron: 0 6 * * * (6:00 UTC = 8:00 Berlin)
- deploy-monitor-erik.sh: one-command deploy to Erik
 2026-03-31 16:52:09 +02:00
Rene Fichtmueller
a952fe3fee feat: ShieldX v0.3.0 — UnicodeScanner (L5), DNS Covert Channel rules, ATLAS v5.4 mappings 
- Layer 4 EntropyScanner: Shannon entropy, Base32/Base64 detection, CVE-2025-55284
  ping/nslookup exfil, EchoLeak markdown pattern, DNS tunneling (iodine/dnscat)
- Layer 5 UnicodeScanner: ASCII Smuggling (U+E0000 Tags Block), Variant Selectors,
  Zero-Width steganography, CamoLeak image-ordering (CVE-2025-53773), homoglyphs,
  BiDi override, high-entropy URL params
- 30 DNS covert channel rules (dns-001 to dns-030)
- ATLASMapper: 29 techniques (ATLAS v5.4.0 Feb 2026), added AML.T0062 (Agent Tool
  Invocation), AML.TA0015 (C2 tactic), memory poisoning, multi-agent trust,
  CamoLeak, Unicode steganography mappings
- Rule count: 72 → 102
- Build: tsup 316ms, zero TypeScript errors
 2026-03-31 16:32:16 +02:00