Hooks Automation

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill forwards arbitrary command strings and file contents into hook commands and persistent logs/memory (e.g., via '${tool.params.command}' and '${tool.params.file_path}'), and enables exporting/logging of commands, files, and session metrics — which can cause secrets to be included verbatim in outputs and logs.

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow ingests user/outsider-authored free text from the ongoing Claude Code tool inputs (e.g., Task description / Bash command / Search query) and passes it into hook commands such as pre-task --description '${tool.params.task}', pre-bash --command '${tool.params.command}', and pre-search --query '${tool.params.pattern}' (all under .claude/settings.json → PreToolUse).