What problem does it solve? Reading a patent's title, abstract, or press coverage tells you almost nothing about what it legally covers — only the claims define the boundary, and misreading them leads to overstated moats or missed freedom-to-operate risks. This Skill applies the examiner's own rules (35 U.S.C. §112 and the MPEP) to dissect a single filing's claim set into a structured, defensible record. ## Core Features & Use Cases - Structural claim parsing: Splits raw claim text into numbered claims, builds the dependency tree, classifies independent versus dependent and multiple dependent claims, and labels each claim's statutory category (method, apparatus, CRM, composition, and more). - Scope determination: Identifies the transitional phrase and states its scope per MPEP §2111.03 — "comprising" is open, "consisting of" is closed — then lists claim 1's elements and names the element a competitor could omit. - Drafting-defect detection: Flags means-plus-function language (§112(f)), relative terms (MPEP §2173.05(b)), negative limitations, Markush groups, and forward, dangling, or improper multiple dependencies, each with its MPEP citation. - Use Case: A competitor announces a patent-pending retrieval product. Paste the published application's claims and receive a structured JSON record: jurisdiction and status from the kind code, claim counts, the open "comprising" transition, claim 1's three elements, the fallback dependent claims, and an Admiralty source grade — with any infringement-shaped observation routed to human review. ## Quick Start Paste the full claim text of a specific patent or published application and ask the agent to analyze what the filing actually covers, including claim 1's breadth and any drafting flags.
