We actively support the following versions of Ananke with security updates:

We take security seriously. If you discover a security vulnerability in Ananke, please report it responsibly.

DO NOT open a public GitHub issue for security vulnerabilities.

Instead, please email security reports to: [email protected]

Include:

• Description of the vulnerability
• Steps to reproduce
• Potential impact
• Suggested fix (if applicable)

• Acknowledgment: Within 48 hours
• Initial Assessment: Within 5 business days
• Status Updates: Every 7 days until resolved
• Resolution Target: Critical issues within 30 days

• We will coordinate disclosure with you
• We request 90 days before public disclosure
• We will credit you in the security advisory (unless you prefer to remain anonymous)

• Store API keys in environment variables
• Use secure credential managers (e.g., 1Password, AWS Secrets Manager)
• Rotate API keys regularly
• Use separate API keys for development and production
• Revoke keys immediately when compromised

• Commit API keys to version control
• Share API keys in plain text
• Use the same API key across multiple environments
• Log API keys in application logs

Example:

# Good: Environment variable
export ANTHROPIC_API_KEY="sk-ant-..."

# Good: From secure credential manager
export ANTHROPIC_API_KEY=$(security find-generic-password -w -s "ananke-api-key")

# Bad: Hardcoded in script
ANTHROPIC_API_KEY="sk-ant-api03-..." # NEVER DO THIS

Set restrictive permissions on configuration files:

# Create config directory
mkdir -p ~/.config/ananke

# Create config file
touch ~/.config/ananke/config.toml

# Set secure permissions (owner read/write only)
chmod 600 ~/.config/ananke/config.toml

Use encrypted storage for sensitive configuration:

# macOS: Use Keychain
security add-generic-password \
  -a "ananke" \
  -s "anthropic-api-key" \
  -w "sk-ant-..."

# Retrieve in scripts
export ANTHROPIC_API_KEY=$(security find-generic-password \
  -a "ananke" \
  -s "anthropic-api-key" \
  -w)

Ananke uses HTTPS for all external API calls:

• Claude API: https://api.anthropic.com
• Modal endpoints: https://*.modal.run

Always verify TLS certificates (enabled by default):

# Verify certificate pinning is enabled
export ANANKE_VERIFY_TLS=true

When using proxies, ensure they support TLS:

# Set HTTPS proxy
export HTTPS_PROXY="https://proxy.example.com:8080"

# DO NOT use HTTP proxies for sensitive traffic
# export HTTP_PROXY="http://proxy.example.com:8080"  # Insecure!

1. Use Official Images:

# Use official Alpine base
FROM alpine:3.19

2. Run as Non-Root User:

# Create non-root user
RUN adduser -D -u 1000 ananke
USER ananke

3. Scan Images for Vulnerabilities:

# Scan with Trivy
trivy image ananke:latest

# Scan with Snyk
snyk container test ananke:latest

4. Limit Container Capabilities:

# docker-compose.yml
services:
  ananke:
    cap_drop:
      - ALL
    security_opt:
      - no-new-privileges:true

5. Use Read-Only Filesystems:

services:
  ananke:
    volumes:
      - ./code:/workspace:ro  # Read-only
    read_only: true

DO:

# Use environment variables
docker run -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY ananke:latest

# Use Docker secrets (Swarm)
docker secret create anthropic_key api_key.txt
docker service create --secret anthropic_key ananke:latest

# Use Kubernetes secrets
kubectl create secret generic ananke-secrets \
  --from-literal=anthropic-api-key=$ANTHROPIC_API_KEY

DON'T:

# Never in Dockerfile
ENV ANTHROPIC_API_KEY="sk-ant-..."  # NEVER DO THIS

Ananke validates all inputs to prevent injection attacks:

File Paths:

• Validated against directory traversal attacks
• Restricted to allowed directories
• Symlink resolution with safety checks

Code Extraction:

• Sandboxed execution for tree-sitter parsing
• No arbitrary code execution
• Resource limits enforced

API Inputs:

• JSON schema validation
• Size limits enforced
• Rate limiting applied

Generated code is sanitized to prevent:

• Command injection
• SQL injection
• XSS attacks
• Path traversal

Never execute generated code without review:

# Bad: Direct execution
eval $(ananke generate "create script")

# Good: Review first
ananke generate "create script" > generated.sh
# Review generated.sh manually
chmod +x generated.sh
./generated.sh

By default, Ananke processes data locally:

• No data sent to external services (unless Claude is enabled)
• All extraction and compilation happens on your machine
• No telemetry or analytics collected

When using Claude for semantic analysis:

• Only code snippets explicitly marked for analysis are sent
• API calls use HTTPS with certificate verification
• No persistent storage of code on Anthropic servers
• Subject to Anthropic's Privacy Policy

When using Modal for generation:

• Only compiled constraints and prompts are sent
• Generated code returned directly to client
• No persistent storage of prompts or outputs
• Subject to Modal's security policies

Disable External Services:

# Extract constraints locally only (no Claude)
ananke extract ./src --no-llm

# Compile locally only
ananke compile constraints.json --no-optimize

Enable audit logging for security-sensitive operations:

# Enable audit log
export ANANKE_AUDIT_LOG="$HOME/.cache/ananke/audit.log"

# Set log level
export LOG_LEVEL="info"

# Rotate logs
logrotate /etc/logrotate.d/ananke

Audit log includes:

• API calls with timestamps
• File access operations
• Constraint compilation events
• Generation requests

Audit log excludes:

• API keys
• Generated code content
• Source code content

Ananke's Zig dependencies are minimal and vendored:

• Standard library only (no external packages)
• All dependencies reviewed before inclusion
• Automated vulnerability scanning via GitHub Dependabot

Maze's Rust dependencies are regularly audited:

# Audit dependencies
cd maze
cargo audit

# Update dependencies
cargo update

# Check for outdated dependencies
cargo outdated

Container base images are regularly updated:

• Alpine Linux security updates applied
• Vulnerability scanning with Trivy/Snyk
• Automated rebuilds on security advisories

# Zig static analysis
zig build test

# Rust static analysis
cd maze
cargo clippy -- -D warnings

# Security-focused linting
cargo clippy -- -D clippy::suspicious

# Run with AddressSanitizer
zig build -Doptimize=Debug -Dsanitize-thread=true

# Rust with sanitizers
cd maze
RUSTFLAGS="-Z sanitizer=address" cargo test

# Scan Rust dependencies
cargo audit

# Scan container images
trivy image ananke:latest

# Scan with Snyk
snyk test

Subscribe to security advisories:

• GitHub: Watch repository → Custom → Security alerts
• Email: [email protected]

# Check current version
ananke --version

# Update to latest version
curl -fsSL https://raw.githubusercontent.com/ananke-ai/ananke/main/scripts/install.sh | bash

# Verify update
ananke --version

Ananke follows security best practices:

• OWASP Top 10: Mitigations implemented
• CWE Top 25: Addressed in design
• NIST Guidelines: Followed where applicable
• Supply Chain Security: SLSA compliance in progress

1. Tree-sitter Parsing: Uses C library with memory safety considerations

   • Mitigation: Sandboxed execution, resource limits
   • Status: Monitoring upstream for security issues

2. Modal Generation: External service dependency

   • Mitigation: HTTPS, input validation, no data retention
   • Status: Production-ready with security best practices

3. Claude API: External service dependency

   • Mitigation: Optional feature, HTTPS, Anthropic's security
   • Status: Production-ready with documented risks

• Code signing for released binaries
• SLSA Level 3 compliance
• Hardware security module (HSM) support for key storage
• Advanced audit logging with tamper detection
• Formal security audit by third party

Before deploying Ananke in production:

• API keys stored securely (not in code/config files)
• Configuration files have restrictive permissions (chmod 600)
• TLS certificate verification enabled
• Running as non-root user (containers and services)
• Audit logging enabled and monitored
• Dependencies up to date (run cargo audit)
• Container images scanned for vulnerabilities
• Input validation tested with edge cases
• Generated code reviewed before execution
• Security updates applied promptly

• OWASP Security Guidelines
• Rust Security Guidelines
• Docker Security Best Practices
• Zig Security

For security concerns:

• Email: [email protected]
• GitHub: @ananke-ai/security-team
• Response Time: Within 48 hours

Last Updated: 2025-11-24 Version: 0.1.0