The Agent Skills Directory

COMMAND_EXECUTION (MEDIUM): The script scripts/search-docs.sh performs shell command execution with unsanitized inputs.
Evidence: The script takes CLI arguments $1 and $2 and uses them directly as $SEARCH_TERM and $CATEGORY inside a grep command. While double-quoted, the lack of input validation allows for argument injection or behavior manipulation of the grep utility.
DATA_EXFILTRATION (MEDIUM): The documentation search tool is vulnerable to Path Traversal.
Evidence: In scripts/search-docs.sh, the variable SEARCH_PATH is constructed by appending the user-supplied $CATEGORY directly to a base path ($DOCS_DIR/guides/$CATEGORY). An attacker could provide a category string containing ../ sequences (e.g., ../../../../etc) to force grep to read sensitive system files outside the designated documentation folder.
DATA_EXFILTRATION (SAFE): The README.md file reveals the author's local system path and username.
Evidence: The installation and usage sections hardcode the path /Users/zach/Documents/cc-skills/docs/supabase/, exposing personal directory structure and a username.
PROMPT_INJECTION (LOW): The skill possesses a surface for indirect prompt injection by ingesting and outputting external documentation data.
Ingestion points: scripts/search-docs.sh reads content from various .txt files on the local filesystem.
Boundary markers: Absent. The script pipes raw file contents to the standard output which is then consumed by the agent without delimiters or instructions to ignore embedded commands.
Capability inventory: The skill can execute shell commands (grep) and access the local filesystem.
Sanitization: None. The skill does not perform any validation or escaping of the content retrieved from the documentation files before presenting it to the agent.

supabase-expert