Skip to content
AI & Agents
Skill

/security-hardening

Harden an application or service against common attack vectors. Use when asked to improve security posture, secure an API, harden a server, or prepare for a security review.

BOOST
From plugin
atlas-agents
58713 skills
Install
$ npx -y skills add agulli/atlas-agents --skill security-hardening --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/security-hardening

Context preview

The summary Claude sees to decide when to auto-load this skill.

Harden an application or service against common attack vectors. Use when asked to improve security posture, secure an API, harden a server, or prepare for a security review.

SKILL.md

security-hardening.SKILL.md
name: security-hardening
description: Harden an application or service against common attack vectors. Use when asked to improve security posture, secure an API, harden a server, or prepare for a security review.
license: MIT

Overview

Security hardening is not about chasing perfect — it's about raising the cost of attack above the attacker's budget. Work systematically through the layers.

Process

Layer 1: Authentication and Authorization

1. Verify all endpoints require authentication. Check for unprotected routes using:

   grep -rn "route\|@app\.\|router\." src/ | grep -v "login\|health\|docs"

2. Check that authorization is enforced at the data layer, not just the route layer. A user who can reach `/orders` should only see *their* orders. 3. Confirm tokens expire. JWTs must have `exp` claims. Sessions must have server-side expiry.

Layer 2: Input Validation

4. Every user-supplied value that touches a database must use parameterized queries. Grep for f-string SQL:

   grep -rn "f\"SELECT\|f'SELECT\|format.*SELECT" src/

5. Every file path from user input must be validated against an allowlist or a sandboxed directory. Check for `os.path.join(base, user_input)` without `os.path.abspath` + `startswith` guard.

Layer 3: Secrets and Configuration

6. Scan for hardcoded secrets:

   grep -rn "password\s*=\s*[\"'][^\"']\|api_key\s*=\s*[\"'][^\"']" src/

7. Confirm all secrets come from environment variables or a secrets manager, not from config files committed to version control. 8. Check `.gitignore` includes `.env`, `*.key`, `*.pem`.

Layer 4: Dependencies and Infrastructure

9. Run the `dependency-audit` skill to check for vulnerable packages. 10. Check HTTP security headers are set (for web services):

  • `Content-Security-Policy`
  • `X-Frame-Options: DENY`
  • `Strict-Transport-Security`
  • `X-Content-Type-Options: nosniff`

Layer 5: Output

11. Write findings as a prioritized list: Critical → High → Medium → Low. For each finding: what it is, where it is (file:line), how to fix it.

Rationalizations

| Excuse | Rebuttal | |--------|----------| | "This is an internal service, it doesn't need hardening" | Internal services are the primary target of lateral movement attacks after initial breach. | | "We'll harden it before the public launch" | Security debt compounds. Fix it before it goes anywhere. | | "The framework handles security" | Frameworks handle some things. You must handle the application-level concerns. |

Verification

  • [ ] All four layers were checked (not just the ones that seemed relevant)
  • [ ] No hardcoded secrets found in source or git history
  • [ ] Every critical and high finding has a specific remediation
  • [ ] A finding was produced for each layer — even if it's "no issues found"
Read more
Ships withatlas-agents

This repository contains the source code, examples, and project implementations for the book "Hands-on AI Agents" — a practical, code-first guide to building AI agents and, more importantly, putting them into production.

Get the whole plugin

Other skills on atlas-agents.