The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation provides installation instructions that involve downloading a shell script from an external URL and piping it directly into the system shell (curl | bash). This executes unverified remote code with the privileges of the current user.
Evidence: curl -fsSL https://squirrelscan.com/install | bash in SKILL.md.
Evidence: irm https://squirrelscan.com/install.ps1 | iex for Windows environments in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to crawl and analyze untrusted external websites. The results of these crawls are fed back to the agent to drive automated code fixes, creating a vulnerability where a malicious website could influence the agent's behavior.
Ingestion points: Untrusted external website content fetched via the squirrel audit command.
Boundary markers: Absent; there are no instructions for the agent to ignore or delimit instructions found within the audited content.
Capability inventory: The agent is granted Bash(squirrel:*) capabilities and is instructed to perform automated file modifications and fixes across the project repository.
Sanitization: None; the skill does not specify any validation or filtering of the audit report content before the agent processes it.
[COMMAND_EXECUTION]: The skill requires the execution of arbitrary shell commands and tool-specific subcommands to perform its tasks.
Evidence: Use of allowed-tools: Bash(squirrel:*) and various squirrel subcommands like init, audit, and report.

audit-website