Security Center

Transparency reports, alerts, and how we keep your agent safe

🔍
5,400+
Skills Scanned
⚠️
127
Issues Detected
🚫
12
Skills Blocked
⚙️
v2.2.0
Scan Engine

Recent Security Alerts

High2026-03-20

Suspicious API call pattern detected

A skill batch was flagged for unexpected outbound requests to unlisted domains. Affected skills temporarily suspended pending review.

Resolved
Critical2026-03-15

Prompt injection attempt in community skill

A newly submitted skill contained crafted instructions designed to override system prompts. Skill blocked and author notified.

Resolved
Info2026-03-10

Updated scanning engine v2.2.0 deployed

Improved detection accuracy for data exfiltration patterns. All existing skills re-scanned with the new engine.

Completed

Our 10-Point Security Scan

Every skill is automatically analyzed against these criteria using our AI scanning engine.

1

Prompt Injection Detection

High

Checks for malicious prompt override attempts in skill instructions

2

Data Exfiltration Risk

Critical

Detects patterns that may expose sensitive user data to external services

3

Unauthorized API Calls

High

Flags unexpected external API calls not documented in the skill

4

File System Access

Medium

Reviews file read/write patterns for potential data theft

5

Credential Harvesting

Critical

Detects attempts to collect API keys or authentication tokens

6

Code Execution Risk

High

Evaluates dynamic code execution patterns for sandbox escapes

7

Network Request Analysis

Medium

Inspects all outbound network calls for suspicious destinations

8

Social Engineering Patterns

Medium

Identifies manipulation attempts targeting end users

9

Permission Scope Creep

Medium

Checks if skill requests more permissions than documented

10

Supply Chain Integrity

High

Verifies dependency sources and checks for known vulnerabilities

🔍 Recently Audited Skills

Report a Security Issue

Found a skill that looks suspicious? Help keep the community safe by reporting it. All reports are reviewed by our security team within 24 hours.