pdd-to-work-order-eval

What problem does it solve?

Work orders generated from PDDs often match the source PDD and decisions perfectly but still fail in real-world use due to unmeasurable verification criteria, off-market commercial terms, or unclear contractual language. This Skill catches these faithful-but-unshippable drafts before they cause rework or contract delays.

Core Features & Use Cases

• 7-dimension LLM-as-Judge grading: Evaluates work orders across contractual clarity, PDD alignment, decisions traceability, verification realism, commercial realism, archetype fit, and writing style, with pass/partial/fail scores and cited evidence for each grade.
• Out-of-chain fitness checks: Explicitly grades verification criteria against real Connect platform capabilities and commercial terms against real Dimagi contracts standards, preventing drafts that are technically PDD-aligned but unenforceable from passing.
• Use case: A contracts team can run this evaluation immediately after a work order is generated from a PDD to catch blocking issues like unmeasurable verification criteria or payment rates below market floor before sending the draft to a partner for review.

Quick Start

Run the pdd-to-work-order-eval skill on the latest work order draft for your Connect opportunity to receive a full quality verdict and a list of any blocking issues that need to be resolved before contract finalization.