Security audit
Security checks for vulnerabilities and agentic risk
This skill appears to convert ebook and document files with Calibre as advertised, with ordinary file-output caution but no evidence of hidden or malicious behavior.
Install this only if you intend to let the agent run Calibre's local ebook-convert command on files you choose. Confirm the input and output paths, avoid sensitive directories, and ask the agent to check before overwriting an existing output file.
The skill invokes shell-based functionality and explicitly instructs running a local script and ebook-convert, but it declares no tool scope such as allowed-tools or permissions. That mismatch can cause the agent to operate with broader-than-necessary execution capability, weakening least-privilege controls and making accidental or unsafe command execution more likely.
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
command = [converter, str(input_path), str(output_path), *normalize_extra_args(args.extra_args)]
try:
completed = subprocess.run(command, check=False)
except OSError as exc:
print(f"[calibre-convert] Failed to start converter: {exc}", file=sys.stderr)
return 4
The skill tells the agent to create output directories and produce output files, but it does not warn about overwriting existing files or other file-modification side effects. In a file conversion workflow this can lead to unintentional data loss or writes to sensitive locations if paths are chosen incorrectly or reused.
No suspicious patterns detected.