DEFEND
AI Runtime Configuration Security
Catch AI-specific misconfigurations before they reach production.
Scan .env files, Docker Compose, Kubernetes manifests, GitHub Actions, Vercel, Supabase, and other config formats for AI-specific unsafe defaults. Detect exposed model APIs, unauthenticated AI runtimes, permissive CORS, debug flags, missing auth, and provider credential risks before they become incidents.
Multi-format
Scan .env, Docker Compose, Kubernetes, GitHub Actions, Vercel, Supabase, and YAML/JSON configs in a single pass.
AI-specific rules
Rules tuned for AI-adjacent risks: exposed model endpoints, unauthenticated AI runtimes, vector DB exposure, and provider credential leaks.
Prioritized findings
Findings ranked by severity with specific line numbers, categories, and remediation guidance.
No execution
Static analysis only — no deployment access required, no production system interaction.
Core capabilities
What AI Config Linter does.
Environment Variable Analysis
Detect secrets in configs, public AI runtime binding, permissive CORS, debug flags, and missing auth across .env and example files.
Container & Compose Review
Flag AI service ports bound to 0.0.0.0, missing auth, no TLS termination, and unsafe default credentials in Docker Compose and Kubernetes.
CI/CD Pipeline Audit
Review GitHub Actions workflows for exposed AI provider keys, unsafe model endpoint usage, and missing secret controls.
Platform Config Review
Check Vercel, Supabase, and platform-specific configs for public AI endpoint exposure and missing access controls.
Prioritized Output
Every finding includes file path, line number, category, severity, evidence, and a specific remediation recommendation.
Evidence-Ready Export
Export findings as JSON for engineering backlog, Jira tickets, or security review evidence.
Evidence & signals
What you get out of the box.
Supported Formats
- .env / .env.example
- Docker Compose
- Kubernetes YAML
- GitHub Actions
- Vercel config
- Supabase config
- Generic JSON/YAML
Risk Categories
- Exposed model APIs
- Unauthenticated AI runtimes
- Permissive CORS
- Debug flags enabled
- Provider credential leaks
- Missing auth
- Vector DB exposure
Deliverables
- Prioritized findings
- Line-level evidence
- Remediation guidance
- JSON export
- Engineering backlog items
AI SECURITY WORKBENCH
Ready to put AI Config Linter to work?
AI Config Linter is an active-development Workbench capability. We review your deployment configuration files and return prioritized findings without executing production workloads.
Continue through the Workbench
Previous MADE stage
Adversarial Range
Scenario-driven AI red-team testing for prompts, agents, tools, RAG, and multimodal systems.
Next MADE stage
Evidence System
Buyer-ready evidence artifacts from AI security assessment and testing.
Related experience
Runtime Trace
Capture and reconstruction of prompts, retrieval, model calls, identities, approvals, tools, outputs, and side effects.
Evidence
Evidence System
Reviewed, buyer-ready evidence artifacts.