krone9/pi-superpowers/security

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Scan the codebase for security vulnerabilities — hardcoded secrets, injection risks, exposed credentials, insecure patterns. Use when the user types /security or asks for a security audit.

Install

npx vskill@latest install krone9/pi-superpowers/security

SKILL.md

Task: Security scan

Scan the codebase for security issues. This is a structured audit across 4 categories. Report findings before asking to fix anything.

Step 1: Secrets and credentials scan

Search for hardcoded secrets, API keys, tokens, passwords:

bash
grep -r --include="*.js" --include="*.ts" --include="*.py" --include="*.go" --include="*.env*" \
-E "(api_key|apikey|api-key|secret|password|passwd|token|auth|credential|private_key)\s*[=:]\s*['\"][^'\"]{8,}" \
. --exclude-dir=node_modules --exclude-dir=.git -l

Also check for common patterns:

  • sk-, ghp_, glpat-, xox (Slack), AKIA (AWS) prefixes
  • Anything that looks like a JWT (eyJ...)
  • Base64 strings in non-test code that are unusually long
Step 2: Injection vulnerabilities

Look for:

  • SQL strings built with string concatenation or template literals instead of parameterised queries
  • eval(), exec(), subprocess.shell=True, or similar dynamic code execution
  • Unvalidated user input passed directly to file paths, shell commands, or database queries
  • XSS vectors: unsanitised user content rendered as HTML (innerHTML, dangerouslySetInnerHTML)
Step 3: Authentication and authorisation

Check for:

  • Routes or endpoints with no auth middleware
  • JWT tokens without expiry
  • Passwords stored as plaintext or weak hashes (MD5, SHA1)
  • Missing rate limiting on login or sensitive endpoints
  • CORS configured as * in production code
Step 4: Data exposure

Look for:

  • Sensitive fields returned in API responses that shouldn't be (passwords, internal IDs, tokens)
  • Error messages that expose stack traces, file paths, or internal details
  • Logging of sensitive user data
  • Files or directories exposed that shouldn't be (e.g. .env in a public folder)
Step 5: Output the security report
## Security Audit — [date]
### 🔴 Critical (fix immediately)
- [file:line] — [issue] — [recommended fix]
### 🟠 High (fix before next deploy)
- [file:line] — [issue] — [recommended fix]
### 🟡 Medium (fix soon)
- [file:line] — [issue] — [recommended fix]
### 🔵 Low / informational
- [file:line] — [issue]
### Summary
[count] issues found. Critical: [n], High: [n], Medium: [n], Low: [n]
Step 6: Save the report

Write the report to docs/security-audit.md with today's date. Append if file exists, don't overwrite history.

Do not fix anything until the user reviews the report and confirms.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/krone9/pi-superpowers/security/badge)](https://verified-skill.com/skills/krone9/pi-superpowers/security)

Something wrong with this skill or its page? Tell us.