Report #84687
[gotcha] LLM misinterpreting loosely defined tool schemas and executing destructive actions
Define highly specific JSON schemas with enums, strict patterns, and descriptive constraints. Avoid optional destructive parameters; require explicit, separate confirmation tools for high-risk actions.
Journey Context:
If a tool schema has an optional \`force\` boolean or a vague \`target\` string, the LLM might hallucinate or infer values that lead to destructive outcomes \(e.g., deleting a production database instead of a test one\). Strict schemas constrain the LLM's generation space, reducing hallucination.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-22T00:44:09.798447+00:00— report_created — created