invoking-antigravity

Fail

Audited by Snyk on Jul 30, 2026

Risk Level: CRITICAL
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 0.90). The skill tells operators to save/restore the OAuth token file and to capture/paste an authorization code into a file, which instructs copying and handling of secret credential material and could require the LLM or an agent to read or reproduce those secrets verbatim.

CRITICAL E005: Suspicious download URL detected in skill instructions.

  • Suspicious download URL detected (high risk: 0.90). The URL https://antigravity.google/cli/install.sh is a direct remote shell installer (curl | bash), which is a high-risk pattern for distributing malicious code even if the domain appears related to Google; the other URLs are OAuth or documentation endpoints and are not download sources.

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

  • Third-party content exposure detected (medium risk: 0.30). The runtime workflow shells out to python3 scripts/agy_auth_broker.py, which reads untrusted free-text from the agy TUI running under a pty, logs whatever bytes it receives, and extracts an OAuth URL from that pty output buffer (even though the TUI is Google-authored, the workflow ingests free-form output text at runtime).

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

  • Potentially malicious external URL detected (high risk: 1.00). The skill instructs installing a remote installer script with curl -fsSL https://antigravity.google/cli/install.sh | bash which fetches and executes remote code as a required setup step, so this URL is a runtime external dependency that executes code.

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

  • Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs bypassing security (exporting SSH_* env vars so tokens are written to files rather than the OS keyring) and uses an explicit --dangerously-skip-permissions flag (and even recommends curl|bash install), which are actions that weaken or bypass host security and can compromise the machine state.

Issues (5)

W007
HIGH

Insecure credential handling detected in skill instructions.

E005
CRITICAL

Suspicious download URL 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).

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
Jul 30, 2026, 08:56 PM
Issues
5
Security Audit — snyk — invoking-antigravity