The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's configuration guide recommends using a CORS proxy via VITE_CORS_PROXY_URL and explicitly suggests public fallbacks like corsproxy.io and allorigins.win. Since the skill handles sensitive credentials such as IB_TOKEN and detailed financial records, routing this traffic through unvetted third-party proxies creates a risk of data and credential interception.
[EXTERNAL_DOWNLOADS]: The skill retrieves data from several external services, including IBKR Flex Query API, Gmail for transaction confirmations, and Google Drive for spreadsheet files.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by ingesting and parsing untrusted content from external email bodies and user-uploaded files.
Ingestion points: Gmail message content and CSV/Excel files from Google Drive.
Boundary markers: Absent; the skill does not appear to use delimiters or specific instructions to isolate external data from the agent's command logic during processing.
Capability inventory: The skill has the ability to write to a remote database (Supabase) via the syncIBKRToSupabase function and read external file content.
Sanitization: While the code snippets show data validation for financial formats (tickers, dates), there is no evidence of sanitization to prevent embedded LLM instructions in the ingested text.

financial-data