The Agent Skills Directory

[SAFE]: The skill uses standard Python libraries (urllib, json, os) to interact with the NexScope API proxy. All network operations target infrastructure managed by the skill's vendor (nexscope-ai).
[INDIRECT_PROMPT_INJECTION]: The skill processes data from the TikTok platform via the Kalodata API. This constitutes a potential surface for indirect prompt injection if external data (such as store names or descriptions) contains instructions meant for the AI agent.
Ingestion points: Data is fetched via scripts/shop_rank.py and scripts/shop_detail.py, then written to local JSON files.
Boundary markers: The instructions do not specify explicit delimiters for the agent when reading the generated files.
Capability inventory: The skill allows the agent to execute Python scripts, write to the filesystem, and use shell tools like jq to read the data.
Sanitization: No explicit sanitization of the API response data is performed before the agent is instructed to read it.
Note: The risk is consistent with standard data retrieval skills and is mitigated by the agent's internal safety guardrails. Findings for this category are consistently rated as low/safe.
[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (scripts/shop_detail.py, scripts/shop_rank.py) and suggests using jq for data processing. These commands are limited to the skill's own functionality and do not involve unsanitized user input in a way that leads to arbitrary command injection.

ecommerce-tiktok-shop-analytics