The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external API responses and user-provided JSON arguments. This creates a surface for indirect prompt injection if the ingested data contains malicious instructions meant to influence the agent's behavior. \n
Ingestion points: sys.argv[1] in various scripts; responses from call_temu_api in _temu_common.py. \n
Boundary markers: None implemented to separate external data from system instructions. \n
Capability inventory: The skill can perform network requests and write to the local file system. \n
Sanitization: Input is validated as JSON, but the content is not sanitized for natural language instructions.\n- [DATA_EXFILTRATION]: The skill transmits business data and access tokens to external endpoints at tool-gateway.linkfox.com and api.linkfox.com. \n
Evidence: Found in scripts/_temu_common.py and scripts/onboarding.py. \n
Note: These domains are owned by the vendor (linkfox-ai) and the communication is essential for the skill's documented functions.\n- [CREDENTIALS_UNSAFE]: The script scripts/onboarding.py contains a hardcoded Base64-encoded identifier used as a fallback for API requests. \n
Evidence: _LOGIN_FIXED_UID = "eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9" (decodes to a JSON object containing an ID). \n
Note: This appears to be a static platform initialization value rather than a sensitive user secret.

linkfox-temu-price-eu