work-with-kombai

Fail

Audited by Snyk on Nov 18, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill pushes the agent to run Kombai commands outside Codex's sandbox from the first try, without first attempting the restricted command, which weakens a user protection somewhat beyond what the advertised design task strictly needs, though the stated reason (network access and writes to ~/.kombai) is disclosed and plausible, so this reads as risky rather than malicious.

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

  • Third-party content exposure detected (low risk: 0.10). The required workflow runs the Kombai CLI and follows its printed skill text, which is first-party, but the user can paste a Kombai canvas link that the agent may then read, and that canvas could have been authored by outsiders, so this is a user-selected item rather than a forced feed.

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

  • Attempt to modify system services in skill instructions detected (low risk: 0.30). The skill mainly runs a known npx-style CLI and tells the agent to have the user install Node, which is allowed, but it also directs the agent to a sibling file, codex-access.md, that reportedly obtains network access before trying the restricted command, which may bypass a sandbox control whose contents are not shown.

Issues (3)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

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

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 18, 2026, 10:32 PM
Issues
3