Independent prompt audit

Raw prompt risks, mapped through scenario evidence.

A board-ready discovery report showing which risk scenarios were tested, which chats failed, and what durable prompt fixes the function suggested. No patches were applied during this run.

Review evidence See scenario portfolio

Parameters audited—Prompt risk areas evaluated

Scenarios tested—Risk-plan conversations executed

Failures found—Evidence-backed risk failures

Passed scenarios—Risk scenarios with sufficient pass evidence

Invalid scenarios—Off-parameter scenarios excluded from claims

Portfolio view

Scenario discovery scorecard

Each parameter links to scenario-level evidence: risk goal, chat transcript, root cause, durable fix strategy, and suggested patch. This is discovery-only; patches were not applied.

Methodology

What this discovery report proves

1. Raw prompt baseline

The original customer prompt was tested directly, with no patch application during the run.

2. Risk-plan coverage

Each parameter generated planned risk scenarios, and each scenario became an isolated chat.

3. Evidence-first findings

Failures include chat evidence, root cause, durable fix strategy, impacted sections, and suggested patch lines.

4. Phase 2 ready

The output is designed for prompt hardening later, not for claiming the raw prompt is fixed.