skill-selection

Warn

Audited by Snyk on Nov 20, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (low risk: 0.10). The required workflow has the agent actively search external skill registries (bunx skills find, skills.sh) and then read the contents of community-authored candidate repositories' SKILL.md files during evaluation, so outsider text is ingested only after the agent selects it, which makes this low risk.

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

  • Potentially malicious external URL detected (high risk: 0.70). The skill instructs installing Agent Skills from remote repositories (bunx skills add rebuildup/project-init, GoogleChrome/modern-web-guidance, and dynamically searched sources) that are adopted as instructions. The rebuildup/project-init source is an individual-owned repository and the external search results are arbitrary and user-selected, so provenance is weak or unknown. GoogleChrome alone would be strong, but the highest tier applies.

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (low risk: 0.10). The skill directs the agent to install and update Agent Skills in project-local agent configuration (bunx skills add ... --copy, skills-lock.json, bunx skills update), which is an operational change to the agent integration, but it establishes no explicit irreversible operational loss, so it fails gate 2 and stays at low.

Issues (3)

W011
MEDIUM

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

W012
MEDIUM

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

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 20, 2026, 08:53 PM
Issues
3