The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a helper script scripts/crawlora.sh to facilitate GET requests to the Crawlora API. The script implements multiple defensive layers, including:
A strict whitelist of allowed API routes (/lululemon/categories, /lululemon/category, etc.).
Path validation using anchored regexes (^/lululemon/product/[^/]+$) to prevent path traversal or route smuggling.
Argument sanitization using curl's --data-urlencode to prevent command injection via query parameters.
Rejection of the @ character in arguments to prevent curl from reading local files.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from api.crawlora.net.
Ingestion points: Data enters the agent context from the Crawlora API response via scripts/crawlora.sh.
Boundary markers: Not explicitly defined in the prompts, though results are processed as JSON.
Capability inventory: The skill can execute shell commands via scripts/crawlora.sh and perform network operations via curl.
Sanitization: External input is URL-encoded before being sent to the API, and the response is expected to be piped into jq for structured parsing, reducing the risk of the agent executing embedded instructions.
[DATA_EXPOSURE]: API credentials are handled safely. The skill requires CRAWLORA_API_KEY to be set in the environment. The helper script writes the key to a temporary, restricted-permission configuration file (using mktemp and chmod 600) which is passed to curl via the --config flag. This prevents the API key from appearing in the system's process table (ps).

lululemon-research