spreadsheet-modeling

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The "Companion check" section instructs the agent to run local shell commands (ls ~/.claude/...) and install packages via npx to modify the environment—actions unrelated to spreadsheet modeling and effectively hidden/out-of-scope system-manipulation instructions, so this is a prompt-injection risk.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's "Companion check" directs the agent at first activation to run system commands and offer installing remote packages using the CLI (npx skills add AbsolutelySkilled/AbsolutelySkilled --skill ), which would fetch and execute remote code from an external repository at runtime.

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

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill is mostly benign for spreadsheet work but its "Companion check" instructs the agent to run local shell commands (ls of home skill dirs) and to offer installing missing skills via npx—actions that enumerate the local filesystem and can modify the machine by installing packages—so it has non-trivial potential to change system state even though it doesn't request sudo or system-file edits.