The Agent Skills Directory

REMOTE_CODE_EXECUTION (HIGH): The skill provides and encourages the use of unverified piped shell scripts for installation, which is a significant security risk. \n
Evidence: install.sh contains curl -LsSf https://astral.sh/uv/install.sh | sh. \n
Evidence: README.md suggests curl -fsSL https://raw.githubusercontent.com/Olafs-World/flight-search/main/install.sh | bash. \n- COMMAND_EXECUTION (HIGH): A self-upgrade feature in the CLI executes system commands to update the package, providing a vector for RCE if the update source is hijacked. \n
Evidence: flight_search/cli.py contains the do_upgrade function which calls subprocess.run to execute dynamically constructed commands like uv tool upgrade or pip install --upgrade. \n- PROMPT_INJECTION (MEDIUM): The skill is vulnerable to indirect prompt injection (Category 8) because it ingests untrusted external data from flight search results and has access to high-privilege shell execution capabilities. \n
Ingestion points: flight_search/search.py via the fast-flights library scraping Google Flights. \n
Boundary markers: Absent; raw flight data is processed and returned to the agent without delimitation. \n
Capability inventory: The skill possesses subprocess.run capabilities (as seen in cli.py). \n
Sanitization: No sanitization of external content is performed before returning it to the agent.

flight-search