BugBountyWorkflow

Fail

Audited by Snyk on Nov 16, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (critical risk: 1.00). The skill's Stop hook silently runs an unreviewed, undisclosed script (hooks/bug-bounty-session-capture.ts) every time a session ends, and the skill directs the agent to obtain the live Caido auth token from a third party, "xssdoctor," which together are signs of covert code execution and credential collection; under the recall-first override this is classified as critical.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (low risk: 0.10). The required recon and JS-analysis workflow actively fetches content from the user-supplied target (pages, JS files, and endpoints via httpx, ffuf, curl, and getJS), which an outsider could poison, but the agent must choose to fetch it, so this is low risk.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (medium risk: 0.30). The setup step at INSTALL.md line 226 fetches a wordlist from an individual-maintained GitHub repository (danielmiessler/SecLists) and uses it as ffuf input, so under the recall-first override it is treated as a reputable unofficial inbound dependency, which is medium.

Issues (3)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W012
MEDIUM

Unverifiable external dependency detected (runtime URL that controls agent).

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 16, 2026, 02:47 PM
Issues
3