The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Etsy and eRank exports, creating a potential surface for indirect prompt injection attacks.
Ingestion points: Data is ingested from external CSV, JSON, and HTML files via functions like load_listings in data_io.py and load_shop_rows in shop_search.py.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded prompts when presenting processed marketplace data to the agent.
Capability inventory: Across its scripts, the skill has the capability to write files (write_csv, write_json), perform network requests via requests, and execute its own CLI module using subprocess (e.g., in run_erank.py).
Sanitization: While reject_sensitive_fields in member_data.py effectively filters for hardcoded credentials or browser session data, there is no validation or sanitization of natural language content in listing fields like title, description, or tags.
[METADATA_POISONING]: The SKILL.md file contains instructional content that promotes the author's social media profile (Douyin ID: 93440780745) for follow-along tutorials and human support.

etsy-tony-full-erank-tools