What problem does it solve? Products often claim "end-to-end encrypted", "zero knowledge", or "not logged", but verifying those claims requires tracing every field through code, storage, logs, and traces. This Skill maps observed fields to those claims and classifies each one with evidence, exposing plaintext leakage that contradicts the stated privacy model. ## Core Features & Use Cases - Field Classification Matrix: Assigns each observed field one of five classes (encrypted_content, encrypted_metadata, plaintext_operational_metadata, avoidable_plaintext_leakage, unknown) with evidence references, confidence, and retest actions. - Multi-Surface Evidence Collection: Guides inspection of schemas, DB migrations, crypto call sites, logs, traces, analytics, HAR captures, and sentinel/differential tests to link the same field across surfaces. - Findings with Severity: Emits findings using a shared schema with PCFC-prefixed ids and severity mappings from critical to informational. - Use Case: While auditing a notes app that claims zero-knowledge encryption, you discover note titles appear in plaintext in server logs and search indexes; the matrix records each field, its readers, and the claim conflict as a high-severity finding. ## Quick Start Ask the agent to map the fields in this repository against the product's end-to-end encryption claim and produce a privacy field matrix with findings.
