The Agent Skills Directory

[SAFE]: The scripts/crawlora.sh helper script implements several security best practices to protect the environment and credentials. It manages the CRAWLORA_API_KEY by writing it to a temporary configuration file with restricted permissions (mode 600), preventing the key from appearing in process command lines, and uses a shell trap to ensure the file is removed on exit. It also enforces a strict allowlist for API routes using anchored regular expressions (e.g., ^/instagram/profile/[^/]+$), preventing access to unintended endpoints.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes untrusted data from external social platforms. 1. Ingestion points: Untrusted data enters the agent context via social media profiles, posts, and comments retrieved from the Crawlora API. 2. Boundary markers: The raw JSON output lacks explicit delimiters or instructions for the agent to ignore embedded commands. 3. Capability inventory: The skill utilizes the Bash tool for network operations and temporary file writes. 4. Sanitization: The script sanitizes its own input parameters but does not perform sanitization on the content returned by the API.
[DATA_EXFILTRATION]: The skill's network traffic is directed to api.crawlora.net. This is a vendor-owned resource associated with the author (crawlora-org) and is used legitimately for the skill's primary research purpose.

social-media-research