zenith-finance-gsheets-sync
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill reads sensitive financial information from local files located at
~/Documents/Obsidian/Notes/Finance/Ledger.mdandTransactions.md. This access is necessary for the primary function of synchronizing data to a spreadsheet. - [COMMAND_EXECUTION]: The skill instructions include the execution of Python scripts for authentication and data processing, such as
~/.hermes/skills/productivity/google-workspace/scripts/setup.pyand~/.hermes/skills/finance/zenith-finance-sync/scripts/sync_wrapper.py. - [PROMPT_INJECTION]: The skill processes untrusted transaction data from external Markdown files, creating a surface for indirect prompt injection.
- Ingestion points:
Ledger.mdandTransactions.mdfiles in the user's Obsidian vault. - Boundary markers: No specific delimiters or instructions are used to isolate the parsed content from the agent's logic during the synchronization process.
- Capability inventory: The skill possesses file read capabilities, shell execution via Python, and write access to the Google Sheets API.
- Sanitization: While the skill utilizes regular expressions to structure the data extraction, it does not specify explicit validation or sanitization of the extracted values before they are sent to the Google Sheets API.
Audit Metadata