tres-recon-gaps
Pass
Audited by Gen Agent Trust Hub on Aug 1, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Data Ingestion and Financial Analysis: The skill processes sensitive financial data, including balances and reconciliation gaps fetched from the TRES Finance MCP connector. This behavior is aligned with the tool's primary purpose and follows expected data handling practices for financial management skills.
- HTML Dashboard Generation: To enhance data visualization, the skill generates a standalone HTML file. This file uses Google Fonts for typography, which is a well-known and trusted service. The instructions emphasize that the dashboard is a pure display interface that does not initiate external network requests or API calls directly from the browser.
- User-Triggered Mutation Workflow: The skill implements an interactive workflow where users can copy specific prompt strings from the HTML dashboard to execute data mutations (e.g., creating plugs or gap-fill rules). While the instructions suggest executing these mutations immediately upon receiving the prompt, the process is gated by the manual user action of copying and pasting the command into the chat session.
- Indirect Prompt Injection Consideration: Because the skill processes data from external TRES Finance queries (such as wallet names or asset identifiers), there is a theoretical surface for indirect prompt injection if those source fields contained malicious instructions. The skill relies on the underlying agent's standard sanitization and safety guardrails when interpreting the data displayed in the dashboard or included in generated prompts.
Audit Metadata