Build reusable case libraries from cross-frame materials

CrossFrame Casebook is a parallel case-library skill to CrossFrame, not a replacement for it - it does not trigger independently, and is only loaded when crossframe-suite routes explicit CrossFrame materials into reusable entries. Its job is narrowly scoped: turn source material into reusable case entries by first protecting facts, sources, and privacy boundaries, then extracting the scale window, mechanism chain, responsibility chain, reverse conditions, reusable concepts, and follow-up observations. The skill body is Chinese-canonical - English is used only for skill IDs, filenames, field names, or external summaries, and Chinese terms govern wherever the two conflict.

Use it only when crossframe-suite routes explicit CrossFrame materials into casebook entries, anonymized case records, mechanism extraction, or retrieval indexes - the goal is future reuse, not immediate advice. Do not use it independently unless the user explicitly names this sibling skill; if the goal shifts to writing up, teaching, debating, or making a public/organizational judgment from an accumulated case, the suite's main dispatcher must be consulted first, since this skill only handles casebook entry structure.

A mandatory sequence: read the base CrossFrame skill for the gates and expression boundaries that apply, consult the routing map to select the right protocol/concept cards/judgment tier for the material's topic, and - if the material triggers high-responsibility, public-institution, intimate-relationship, long-term-evolution, framework-governance, AI-reality-verification, weak-signal/opaque, no-exit, instrumentalization, or metaphor/source-transparency conditions - pull in the continuity bundles and run a source-continuity worksheet (skipping this means the output must be downgraded). It reuses the base skill's read-state-capsule template rather than reinventing source routing, applies material-boundary protocols to separate source/fact/speculation/privacy/publishability layers, and a casebook-build protocol to decide whether the task is a new case, cleaning an old one, batch indexing, comparing cases, or converting a retrospective into a case. A field guide ensures every case records at least nine elements: case summary, factual boundary, material sources, scale window, mechanism chain, responsibility chain, reverse conditions, reusable concepts, and follow-up observations - with privacy/redaction rules applied to names, organizations, locations, timestamps, raw chat text, screenshots, links, and other identifying detail. A mechanism-extraction protocol specifically prevents the output from being just narrative or a pile of concepts without an actual causal chain.

Hard rules: never copy the full CrossFrame skill text (only relative-path reads of the canonical skill and routing map); never turn raw chat text or personal information directly into a case asset unless the user explicitly requests it within a confirmed publishable scope; never write speculation, motive inference, or secondhand opinion as fact; every case must extract at least one mechanism chain and one responsibility chain, not just tell a story; never let CrossFrame terminology substitute for case facts - concepts must serve reuse, not decorate output; never let a casebook become a character trial, organizational conviction, public verdict, or compliance endorsement; and when evidence is insufficient but risk is urgent, only low-risk, revocable, observable follow-up items may be output.

Users working within the CrossFrame suite who need source material distilled into a reusable, privacy-safe, structurally-grounded case library entry rather than a one-off narrative or advice response.