VibeSec 🛡️
Autonomous AI & Application Security Platform: SAST, Secrets, Polyglot SCA, Air-Gapped ONNX False-Positive Noise Triage, and Verified Auto-Remediation — engineered by CyberEiche under the CyberEiche security ecosystem alongside cybereiche-mcp-sec.
VibeSec is an enterprise-grade, privacy-first security scanner built for modern development teams. It detects vulnerabilities, hardcoded secrets, polyglot dependency CVEs, and AI/LLM safety risks (OWASP Top 10 for LLMs), automatically triaging false positives with a built-in, air-gapped INT8 ONNX transformer model.
🏛️ DACH & EU Enterprise Security & Compliance
VibeSec is designed specifically to comply with European Union and DACH regulatory frameworks (Germany, Austria, Switzerland), ensuring enterprise security teams can adopt it without data sovereignty risks or legal hurdles:
- Strict Air-Gap & Zero Cloud Telemetry (GDPR & Data Sovereignty):
- Scans and machine-learning noise triage execute 100% locally in ephemeral process memory.
- Zero source code, snippets, or metadata leave your local machine or internal GitHub Actions runner (0 outbound network bytes, 0 background telemetry).
- Formal architectural compliance documented in DACH_EU_COMPLIANCE_AND_THREAT_MODEL.md.
- EU AI Act Ready (Regulation 2024/1689):
- Complies with Articles 9–15 (Chapter III High-Risk requirements): robust risk management, technical documentation, record-keeping, and human-in-the-loop oversight.
- Every AI-driven triage decision includes full explainability, calibrated confidence scores ($P(\text{Exploitable})$), and structured property bags in OASIS SARIF v2.1.0 exports.
- Asymmetric Zero-Drop Safety Policy:
- Critical vulnerabilities (CVSS $\ge 9.0$) and Remote Code Execution (RCE) patterns are never auto-suppressed, guaranteeing zero masking of catastrophic risks.
- Automated European PII Redaction (GDPR Article 25 - Privacy by Design):
- Context harvesting features built-in, linear-time ($< 2\text{ ms}$) redaction of European IBANs, German Tax IDs (Steueridentifikationsnummer), mobile numbers, emails, and credentials prior to model tokenization.
- Supply-Chain Integrity & Tamper Protection (NIS2 Directive & BSI IT-Grundschutz):
- Compliant with BSI IT-Grundschutz modules
APP.6,CON.2,OPS.1.1.4, andSEC.1. - The compiled INT8 ONNX model is verified at load time using Ed25519 digital signatures against the CyberEiche Master Public Key and streaming SHA-256 chunked digest validation. Any altered or unverified weight is instantly rejected (fail-closed).
- Compliant with BSI IT-Grundschutz modules
🚀 Key Features
- Local INT8 AI Noise Triage:
- In-memory, compiled transformer model (22.08 MB) executing dynamic INT8 inference in ~15–20 ms on pure CPU.
- Cuts static analysis alert fatigue by 40–60% by filtering benign idioms, dead-code branches, and test harness mocks.
- Unified Scanning Quadriad:
- SAST: Multi-language static analysis with custom rulesets and Semgrep compatibility.
- Secrets: High-entropy credential and API token detection (Gitleaks & Yelp
detect-secrets). - Polyglot SCA: Zero-runtime lockfile scanning powered by Google OSV (
osv.dev) across Python, JS/TS, Go, Rust, Java, PHP, Ruby, and .NET. - AST & LLM Security: Specialized AST patterns for Prompt Injection, agent shell execution, and unvalidated LLM outputs.
- Target Truth Validation:
- Reachability-driven active verification layer (DNS → TCP → HTTP) ensuring zero-noise and confirmed endpoint reachability.
- Zero-Setup GitHub Actions:
- Native PR annotations (
::error::),$GITHUB_STEP_SUMMARYaudit scorecards, and OASIS SARIF v2.1.0 output for the GitHub Security tab without third-party bot tokens.
- Native PR annotations (
- Autonomous Verified Remediation:
- Synthesizes localized patch diffs and verifies them against syntax checks and automated test suites before suggesting changes.
📦 Installation
pip install cybereiche-vibesec
Requirements: Python 3.10, 3.11, or 3.12 (macOS, Linux, Windows).
🛠️ User Guide & CLI Usage
1. Basic Scan (Default with AI Noise Triage)
Scans the current repository, runs all scanners in parallel, and applies local ONNX noise reduction:
vibesec scan .
2. Inspect Filtered False Positives
View a dedicated breakdown of findings suppressed by AI noise reduction, including confidence scores and justification reasons:
vibesec scan . --show-suppressed
3. Strict Compliance Mode
For high-assurance or regulated environments, elevated sensitivity flags all borderline findings for human review:
vibesec scan . --strict
4. Raw Scanner Mode (Disable AI Triage)
Bypasses the ONNX triage engine and displays raw, unfiltered scanner output:
vibesec scan . --no-triage
5. Export OASIS SARIF Report (CI/CD Pipelines)
Generates schema-compliant SARIF v2.1.0 including triage suppression objects for the GitHub Security tab or GitLab Security Dashboard:
vibesec scan . --sarif vibesec-results.sarif --severity HIGH
6. Autonomous Verified Remediation
Preview or apply automated code patches:
# Preview fixes (dry-run)
vibesec fix . --dry-run
# Apply verified fixes in-place only if pytest / npm test passes
vibesec fix . --apply --run-tests
🤖 Remediation Configuration (Optional)
VibeSec synthesizes localized security patches verified against AST syntax checks and your local test suite. Powered by LiteLLM, it supports Anthropic, OpenAI, DeepSeek, and self-hosted local models (Ollama, vLLM).
Configure credentials via .env:
# 1. Anthropic (Claude - Default)
ANTHROPIC_API_KEY="sk-ant-..."
# 2. OpenAI (GPT-4o)
OPENAI_API_KEY="sk-proj-..."
# 3. Local Air-Gapped Gateway (Ollama, vLLM)
VIBESEC_LLM_MODEL="ollama/llama3"
VIBESEC_LLM_API_BASE="http://localhost:11434/v1"
Or configure via .vibesec.yaml:
remediation:
enabled: true
model: "claude-3-5-sonnet-20241022" # or gpt-4o, ollama/llama3
⚙️ Project Configuration (.vibesec.yaml)
Customize scanner and triage settings by creating .vibesec.yaml in your project root:
version: 1
settings:
severity_threshold: "HIGH" # LOW | MEDIUM | HIGH | CRITICAL
fail_on_findings: true
triage:
enabled: true
threshold_confirm: 0.65
threshold_suppress: 0.15
zero_drop_critical: true
scanners:
sast:
enabled: true
secrets:
enabled: true
sca:
enabled: true
ignore:
paths:
- "tests/**"
- "docs/**"
- "node_modules/**"
🛡️ GitHub Actions Integration
Add VibeSec to .github/workflows/security.yml:
name: VibeSec Security & AI Guardrails
on:
push:
branches: [main]
pull_request:
branches: [main]
jobs:
security-audit:
runs-on: ubuntu-latest
steps:
- name: Checkout Repository
uses: actions/checkout@v4
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: "3.11"
- name: Install VibeSec
run: pip install cybereiche-vibesec
- name: Run VibeSec Security Audit
run: vibesec scan . --severity HIGH --sarif vibesec-results.sarif
- name: Upload SARIF to GitHub Security Tab
if: always()
uses: github/codeql-action/upload-sarif@v3
with:
sarif_file: vibesec-results.sarif
📄 License
VibeSec is distributed under the Apache 2.0 License.
Developed and maintained by CyberEiche.
🌲 About CyberEiche
CyberEiche is an enterprise cybersecurity firm based in Germany, specializing in autonomous AI defense, application security, and protocol verification.
- Website: cybereiche.de
- Contact: contact@cybereiche.de
- Ecosystem: cybereiche-mcp-sec • cybereiche-vibesec
Metadata
Release files for cybereiche-vibesec 0.2.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| cybereiche_vibesec-0.2.1.tar.gz | 18.7 MB | Without provenance | Details for cybereiche_vibesec-0.2.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| cybereiche_vibesec-0.2.1-py3-none-any.whl | Python 3 | none | any | Details for cybereiche_vibesec-0.2.1-py3-none-any.whl |
Total release size: 37.4 MB
Release files / cybereiche_vibesec-0.2.1.tar.gz
| Download URL | cybereiche_vibesec-0.2.1.tar.gz |
|---|---|
| Size | 18.7 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8d9b17444df7a8a1ba8b33e4b11435417764902e359203af32842b584e11373f
|
|
BLAKE2b-256 checksum How to use checksums |
b3eb06fec5f13eae59db82131734ab199f310f6d04b37162cea4849876c88882
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.11.15
|
Release files / cybereiche_vibesec-0.2.1-py3-none-any.whl
| Download URL | cybereiche_vibesec-0.2.1-py3-none-any.whl |
|---|---|
| Size | 18.7 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
adc3a30ac93558e84fb2f2f6f1228f2f7edacec41d716dea103ae9086ff6ce8d
|
|
BLAKE2b-256 checksum How to use checksums |
347f752233d412843e9efb4fb2f463587db90cf7043c9a8381bd8cee7a77335b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.11.15
|
Security for cybereiche-vibesec 0.2.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.