The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command scripts/handle_technical_anomaly.py <STANDARD_SYMBOL> ... where the stock symbol and indicator filters are derived from user input. Although the skill mandates a normalization step, this pattern is susceptible to command injection if the normalization is bypassed or if the user provides input specifically designed to break out of the shell command context (e.g., using semicolons or backticks).
[PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection as it processes untrusted user data which is eventually passed to a shell-executed script.
Ingestion points: User-provided stock mentions and technical indicator requests in SKILL.md.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to ignore embedded commands within the user input.
Capability inventory: Subprocess execution of scripts/handle_technical_anomaly.py via the system shell.
Sanitization: Relies on the AI agent's ability to normalize user input to a fixed set of standard symbols (e.g., US.NVDA), which is a logic-based validation rather than a robust sanitization boundary.
[REMOTE_CODE_EXECUTION]: The script scripts/handle_technical_anomaly.py uses dynamic path modification (sys.path.insert) to load the 'futu' library from the local filesystem. This dynamic loading mechanism could be exploited to execute arbitrary code if a malicious package is placed in the expected directory path.

moomoo-technical-anomaly