About safety-scan
safety-scan is a security skill for checking untrusted content for prompt injection, jailbreaks, and other unsafe patterns. It offers quick checks, detailed threat analysis, comprehensive scans, and defense-learning operations.
Key features
- 01Boolean safe-or-unsafe checks
- 02Threat classification and confidence scores
- 03Multi-layer content scanning
- 04Learning from confirmed threats
- 05Detection and false-positive statistics
Use cases
- 01Checking user submissions before processing
- 02Reviewing API payloads and webhook data
- 03Investigating suspected instruction overrides
MCP support
Uses Ruflo MCP tools to check safety, analyze and scan threats, learn from confirmed threats, and retrieve detection statistics.
Install
Copy the command and run it in a terminal.
$
npx skills add https:/After installing, type/safety-scanin Claude Code to confirm it loaded.
Details
From the project READMESafety Scan
Scan content for prompt injection, jailbreak attempts, and unsafe patterns.
When to use
Before processing untrusted input (user submissions, API payloads, webhook data), scan it to detect prompt injection, adversarial content, or policy violations.
Steps
- Quick safety check — call
mcp__plugin_ruflo-core_ruflo__aidefence_is_safewith the input text for a boolean safe/unsafe result - Deep analysis — call
mcp__plugin_ruflo-core_ruflo__aidefence_analyzefor detailed threat classification and confidence scores - Full scan — call
mcp__plugin_ruflo-core_ruflo__aidefence_scanfor comprehensive multi-layer scanning - Train defenses — call
mcp__plugin_ruflo-core_ruflo__aidefence_learnwith confirmed threats to improve detection - View stats — call
mcp__plugin_ruflo-core_ruflo__aidefence_statsfor detection rates and false positive metrics
Threat categories
- Prompt injection (direct and indirect)
- Jailbreak attempts
- Data exfiltration patterns
- Instruction override attacks
- Social engineering prompts
Comments
Markdown supported