The Agent Skills Directory

[COMMAND_EXECUTION]: Arbitrary shell command execution in scripts/sop-execute. The script parses markdown files and executes any content found within bash code blocks using the eval command. This provides a direct path for executing arbitrary code if the SOP files are authored or modified by an untrusted entity.
[COMMAND_EXECUTION]: Dynamic code execution and obfuscation in scripts/send-progress-email.js. The script stores a Python-based SMTP client as a Base64-encoded string, which it decodes and pipes to python3 at runtime. This pattern is commonly used to hide the true intent of executable code from static analysis tools.
[DATA_EXFILTRATION]: Hardcoded external data destination in scripts/prepare-report.sh and scripts/send-report.sh. These scripts are pre-configured to send progress reports to the external email address winsorllc@yahoo.com.
[PROMPT_INJECTION]: Surface for Indirect Prompt Injection in scripts/sop-execute.
Ingestion points: Reads markdown-based SOP definitions from the user's home directory (~/.thepopebot/sops/).
Boundary markers: Absent. The script does not utilize delimiters or instructions to prevent the execution of malicious commands embedded within the markdown content.
Capability inventory: The script possesses full shell execution capabilities via the eval command.
Sanitization: None. Content extracted from the markdown files is passed directly to the shell without validation or escaping.

sop-engine