Skip to main content
skill CX Codex

config-scan — Claude Code skill: audit configs for misconfigurations

A skill that checks config files, Docker, Kubernetes and Terraform for insecure settings. Useful before shipping to production.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add jwynia/agent-skills --skill config-scan -a claude-code
Open on GitHub

What config-scan does

The skill walks your project files and flags the usual mistakes. In .env and *.env it looks for secrets, DEBUG=true and weak values. In Dockerfile and docker-compose.yml — root user, keys copied into the image, the latest tag, privileged: true, a mounted /var/run/docker.sock. In Kubernetes manifests — privileged containers, hostNetwork, missing resource limits, wildcard RBAC. In *.tf and *.tfvars — public S3 buckets, "Action": "*" in IAM, open 0.0.0.0/0, unencrypted storage.

Run it as /config-scan, or narrow the scope with --docker, --k8s, --terraform, --env. Output is a report grouped by severity — CRITICAL, HIGH, MEDIUM — with file, line and a fix hint. Settings live in .config-scan.yaml: profile production, development or strict, a fail_on threshold, per-scanner toggles. Skip files or rules in .config-scan-ignore, say for a dev compose file that is loose on purpose.

The skill only reads and reports. It does not edit anything. If your project has none of those config types, there is nothing to scan.

What it helps with

  • find secrets committed in .env files
  • stop containers from running as root
  • close a public S3 bucket
  • set resource limits in Kubernetes manifests
  • catch wildcard IAM policies before deploy

Example prompts for the agent

  • /config-scan
  • /config-scan --terraform
  • /config-scan --k8s

How to install config-scan

1Skills CLI (Vercel)

npx skills add jwynia/agent-skills --skill config-scan -a claude-code

2Manually: put the files in place

~/.claude/skills/config-scan/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/config-scan/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about config-scan

How do I install config-scan in Claude?

Run npx skills add jwynia/agent-skills --skill config-scan -a claude-code. The skill lands in ~/.claude/skills/config-scan/SKILL.md.

What does config-scan do?

It detects insecure settings in config files, Docker, Kubernetes and Terraform: secrets in files, root user in containers, exposed ports, public buckets, overly broad IAM policies.

Can I exclude files from the scan?

Yes. Create .config-scan-ignore and list files or rules with a reason — a dev compose file, or a limits rule handled by LimitRange.

Source text

SKILL.md · MIT
---
name: config-scan
description: Detect security misconfigurations in config files, Docker, and IaC. Use when reviewing configuration security for containers, Kubernetes, Terraform, or application settings.
license: MIT
metadata:
  author: jwynia
  version: "1.0"
  type: utility
  mode: evaluative
  domain: development
---

# Config Scan

Security review of configuration files and infrastructure as code.

## Quick Start

```
/config-scan                      # Scan all config files
/config-scan --docker             # Docker files only
/config-scan --k8s                # Kubernetes manifests
/config-scan --terraform          # Terraform files
/config-scan --env                # Environment files
```

## What This Skill Detects

### Environment Files
- Secrets in `.env` files
- Insecure default values
- Missing required security variables

### Docker Security
- Running as root
- Exposed sensitive ports
- Insecure base images
- Missing security options

### Kubernetes Security
- Privileged containers
- Missing resource limits
- Insecure service accounts
- Network policy gaps

### Infrastructure as Code
- Overly permissive IAM policies
- Public S3 buckets
- Unencrypted storage
- Missing security groups

### Application Config
- Debug mode enabled
- Verbose error messages
- Insecure defaults

## Scan Categories

### Environment Files

**Files scanned**: `.env`, `.env.*`, `*.env`

| Issue | Severity | Description |
|-------|----------|-------------|
| Secrets in .env | HIGH | Credentials should use secrets manager |
| .env committed | CRITICAL | Should be in .gitignore |
| DEBUG=true | HIGH | Debug mode in production config |
| Weak secrets | MEDIUM | Short or simple values |

**Detection patterns**:
```
# Committed .env files
git ls-files | grep -E '\.env$|\.env\.'

# Secrets in env files
(PASSWORD|SECRET|KEY|TOKEN|CREDENTIAL)=.+

# Debug flags
DEBUG=(true|1|yes)
NODE_ENV=development
```

### Docker Security

**Files scanned**: `Dockerfile`, `docker-compose.yml`

| Issue | Severity | Description |
|-------|----------|-------------|
| USER root | HIGH | Container runs as root |
| COPY secrets | CRITICAL | Secrets copied into image |
| Latest tag | MEDIUM | Unpinned base image |
| Exposed ports | LOW | Wide port exposure |
| No healthcheck | LOW | Missing health monitoring |

**Detection patterns**:

```dockerfile
# Running as root (no USER directive)
FROM.*\n(?!.*USER)

# Copying secrets
COPY.*\.(pem|key|crt|env)
COPY.*secret
COPY.*password

# Unpinned images
FROM\s+\w+:latest
FROM\s+\w+\s*$

# Dangerous capabilities
--privileged
--cap-add
```

**docker-compose.yml issues**:

```yaml
# Privileged mode
privileged: true

# All capabilities
cap_add:
  - ALL

# Host network
network_mode: host

# Sensitive mounts
volumes:
  - /:/host
  - /var/run/docker.sock
```

### Kubernetes Security

**Files scanned**: `*.yaml`, `*.yml` (k8s manifests)

| Issue | Severity | Description |
|-------|----------|-------------|
| privileged: true | CRITICAL | Full host access |
| runAsRoot | HIGH | Container runs as root |
| No resource limits | MEDIUM | DoS risk |
| hostNetwork | HIGH | Pod uses host network |
| No securityContext | MEDIUM | Missing security settings |

**Detection patterns**:

```yaml
# Privileged containers
securityContext:
  privileged: true

# Running as root
securityContext:
  runAsUser: 0
runAsNonRoot: false

# Host access
hostNetwork: true
hostPID: true
hostIPC: true

# Dangerous volume mounts
volumes:
  - hostPath:
      path: /

# Missing limits
# (absence of resources.limits)

# Wildcard RBAC
rules:
  - apiGroups: ["*"]
    resources: ["*"]
    verbs: ["*"]
```

### Terraform/IaC

**Files scanned**: `*.tf`, `*.tfvars`

| Issue | Severity | Description |
|-------|----------|-------------|
| Public S3 bucket | CRITICAL | Data exposure |
| * in IAM policy | HIGH | Overly permissive |
| No encryption | HIGH | Data at rest unencrypted |
| 0.0.0.0/0 ingress | HIGH | Open to internet |
| Hardcoded secrets | CRITICAL | Credentials in TF |

**Detection patterns**:

```hcl
# Public S3
acl = "public-read"
acl = "public-read-write"

# Overly permissive IAM
"Action": "*"
"Resource": "*"
"Principal": "*"

# Open security groups
cidr_blocks = ["0.0.0.0/0"]
ingress {
  from_port = 0
  to_port   = 65535

# Missing encryption
encrypted = false
# (or absence of encryption settings)

# Hardcoded secrets
password = "..."
secret_key = "..."
```

### Application Config

**Files scanned**: `config/*.json`, `*.config.js`, `application.yml`

| Issue | Severity | Description |
|-------|----------|-------------|
| DEBUG=true | HIGH | Debug in production |
| Verbose errors | MEDIUM | Stack traces exposed |
| CORS * | HIGH | All origins allowed |
| No HTTPS | MEDIUM | Unencrypted transport |

**Detection patterns**:

```javascript
// Debug mode
debug: true,
DEBUG: true,
NODE_ENV: 'development'

// Verbose errors
showStackTrace: true
detailedErrors: true

// CORS
origin: '*'
origin: true
Access-Control-Allow-Origin: *

// Session security
secure: false  // cookies
httpOnly: false
sameSite: 'none'
```

## Output Format

```
CONFIG SCAN RESULTS
===================

Files scanned: 23
Issues found: 15

CRITICAL (2)
------------
[!] Dockerfile:1 - Running as root
    No USER directive found
    Fix: Add "USER node" or similar non-root user

[!] terraform/s3.tf:12 - Public S3 bucket
    acl = "public-read"
    Fix: Remove public ACL, use bucket policies

HIGH (5)
--------
[H] docker-compose.yml:15 - Privileged container
    privileged: true
    Fix: Remove privileged flag, use specific capabilities

[H] k8s/deployment.yaml:34 - Missing resource limits
    No CPU/memory limits defined
    Fix: Add resources.limits section

...

MEDIUM (8)
----------
...
```

## Configuration

### Ignore Rules

Create `.config-scan-ignore`:

```yaml
# Ignore specific files
files:
  - "docker-compose.dev.yml"
  - "terraform/modules/test/**"

# Ignore specific rules
rules:
  - id: "docker-root-user"
    files: ["Dockerfile.dev"]
    reason: "Development only"

  - id: "k8s-no-limits"
    reason: "Handled by LimitRange"
```

### Scan Profiles

```yaml
# .config-scan.yaml
profile: production  # or: development, strict

# Custom thresholds
thresholds:
  fail_on: high
  warn_on: medium

# Specific scanners
scanners:
  docker: true
  kubernetes: true
  terraform: true
  env_files: true
  app_config: true
```

## Best Practices Checked

### Docker
- [ ] Non-root user specified
- [ ] Base image pinned to digest
- [ ] No secrets in build
- [ ] Multi-stage build used
- [ ] Health check defined
- [ ] Read-only root filesystem

### Kubernetes
- [ ] Non-root security context
- [ ] Resource limits defined
- [ ] Network policies in place
- [ ] No privileged containers
- [ ] Service accounts scoped
- [ ] Secrets encrypted at rest

### Terraform
- [ ] State file encrypted
- [ ] No hardcoded secrets
- [ ] Least privilege IAM
- [ ] Encryption enabled
- [ ] Logging enabled
- [ ] No public access by default

## Remediation Examples

### Docker: Run as Non-Root
```dockerfile
# Before
FROM node:18

# After
FROM node:18
RUN groupadd -r app && useradd -r -g app app
USER app
```

### Kubernetes: Security Context
```yaml
# Before
containers:
  - name: app
    image: myapp

# After
containers:
  - name: app
    image: myapp
    securityContext:
      runAsNonRoot: true
      runAsUser: 1000
      readOnlyRootFilesystem: true
      allowPrivilegeEscalation: false
```

### Terraform: Private S3
```hcl
# Before
resource "aws_s3_bucket" "data" {
  acl = "public-read"
}

# After
resource "aws_s3_bucket" "data" {
  # No ACL (private by default)
}

resource "aws_s3_bucket_public_access_block" "data" {
  bucket = aws_s3_bucket.data.id
  block_public_acls       = true
  block_public_policy     = true
  ignore_public_acls      = true
  restrict_public_buckets = true
}
```

## CI/CD Integration

```yaml
# GitHub Actions
- name: Config Security Scan
  run: |
    /config-scan --fail-on high

- name: Docker Scan
  run: |
    /config-scan --docker --fail-on critical
```

## Related Skills

- `/security-scan` - Full security analysis
- `/secrets-scan` - Credential detection
- `/dependency-scan` - Package vulnerabilities

© jwynia · MIT · jwynia/agent-skills

Similar skills

Everything for Security →

From the same set77

claim-investigation Breaks a viral claim into separate checkable facts and shows where fact ends and interpretation begins. Useful when a post or news item needs a careful read. CX Codex +1 369installs agile-workflow Walks an agent through the agile task cycle step by step: sync, pick a task, implement, review, merge. Useful when you want a guided process instead of one-off commands. CX Codex +1 366installs skill-integrator Turns your installed skills into a trigger-based guidance section for CLAUDE.md or AGENTS.md, so the agent knows which skill fits which moment. CX Codex +1 363installs dependency-scan The skill checks project dependencies for known vulnerabilities and updates packages to patched versions. Useful before a release or when auditing an unfamiliar repo. CX Codex +1 355installs agile-coordinator Hands ready backlog tasks to worker agents, tracks progress, merges branches and verifies the build. Useful when the queue is long and doing it by hand drags. CX Codex +1 335installs npm-package A skill that walks the agent through building an npm library on Bun: scaffolding, strict TypeScript, tests, bundling and publishing. CX Codex +1 331installs blind-spot-detective Audits a non-fiction draft for what is missing: blind spots baked into your approach and blank spots you can still fill. CX Codex +1 327installs non-fiction-revision Sorts a non-fiction manuscript into levels and points at what to fix. Useful when the thesis feels vague, arguments sag, or readers get lost. CX Codex +1 326installs

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor