dotnet-csrun

Fail

Audited by Snyk on Nov 21, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill's remote-release execution is openly disclosed and serves the advertised csrun purpose, but the native block runs whatever the latest release is each time, with a checksum taken from the same GitHub API response, which is a weakened protection beyond what a pinned run would need.

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

  • Third-party content exposure detected (low risk: 0.10). The required workflow reads outsider-authored text only when the agent fetches published NuGet package metadata and package documentation to verify exact pins, which is a public registry the agent must actively query and does not come from a monitored queue.

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

  • Potentially malicious external URL detected (medium risk: 0.30). The generated native blocks fetch the latest codebeltnet/csrun csrun.ps1 release asset at runtime and execute it, so this is an inbound dependency; the publisher is an established organization, but the only check is a same-channel digest on a mutable latest release, so it is not independently authenticated and is classified as reputable unofficial (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 21, 2026, 05:09 PM
Issues
3