PromptLockr

PromptLockr is a web app for teams shipping LLM features who need predictable, testable outputs. You store “approved” prompts as versioned test cases with locked model/provider settings, system prompts, tools, and expected responses (or graded rubrics). On a schedule or on every deployment, PromptLockr re-runs the suite against the current model versions and compares results using deterministic checks (exact match/JSON schema) plus semantic scoring and policy checks. When drift is detected—model updates, temperature changes, tool output shifts, or prompt edits—it flags the regression, shows a diff, and explains which constraint failed. Teams can set thresholds to alert, open a ticket, or block CI merges. Audit trails, approvals, and exportable reports help with compliance and incident reviews. Built for web-first workflows, with CI/CD integrations and a lightweight API for programmatic runs.