The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to download and execute an installation script (install.sh) directly from the vendor's GitHub repository using a curl | bash pipeline.
Evidence: curl -sfL https://raw.githubusercontent.com/ghostsecurity/reaper/main/scripts/install.sh | bash in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to capture, inspect, and display raw HTTP/HTTPS traffic from external sources, creating a surface for indirect prompt injection.
Ingestion points: Commands such as reaper logs, reaper search, and reaper get <id> read and display untrusted data from captured network requests and responses.
Boundary markers: The instructions do not define delimiters or warnings to treat captured traffic as untrusted content.
Capability inventory: The agent can execute arbitrary shell commands via the CLI, providing a path for exploitation if the agent obeys instructions embedded in captured traffic.
Sanitization: There is no evidence of filtering or sanitization of the captured HTTP bodies or headers before they are presented to the agent.
[COMMAND_EXECUTION]: The skill uses the reaper CLI tool to manage a local proxy daemon and interact with a SQLite database (~/.reaper/reaper.db).

ghost-proxy