菜单
Use when reviewing or publishing approved LEV candidate bundles. Compares staged JSON to live records and evidence-audit mappings, then uses the file-backed publish workflow.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when revising the public current LEV story, homepage State of LEV story, what to watch next, where better evidence is needed, or story revision triggers. Compares the current story against public outlooks, state-of-the-field editions, public updates, and review due dates without introducing new unreviewed evidence.
Use when coordinating agent-led LEV reconciliation across public updates, State of the Field editions, current LEV story, hallmark insights, track pages, and editorial workflow manifests. Produces structured agent recommendations, asks the curator for approval or revision only when meaningful, then executes approved work through the specialist LEV skills and verification commands.
Use when creating or revising LEV State of the Field editions, especially monthly public updates that need to explain what changed, what did not change, why it matters, and whether the evidence makes LEV look closer without introducing unreviewed evidence.
Use when checking what changed since the last LEV review for one scope. Compares public records, evidence audit trails, PubMed, registries, and other current external sources, then stages JSON plus a surveillance bundle without publishing.
Use when initializing LEV coverage for one hallmark, track, or intervention family. Stages source/study/finding/outlook evidence-map JSON plus a candidate bundle without publishing.
Use when verifying a LEV candidate bundle. Checks source fidelity, interpretation, and evidence-audit mappings; writes evidence-review records without publishing.
| name | lev-editorial-review |
| description | Use when reviewing or publishing approved LEV candidate bundles. Compares staged JSON to live records and evidence-audit mappings, then uses the file-backed publish workflow. |
Use this for the admin side of the tracker after research has already produced a candidate bundle.
data/candidate-bundles/<bundle-id>.jsondata/staged-records/<bundle-id>/file_pathdata/review-comments/ and data/publication-events/research/coverage-assessments/ when a track outlook or support map changesscripts/research-bundle.mjssrc/lib/site-data.ts and src/app/admin/review/npm run research:bundle -- status --bundle <bundle-id> and inspect lifecycle status, proposed changes, prior comments, and the evidence-review gate.npm run research:bundle -- validate --bundle <bundle-id> before any status change.record_typeidfile_pathstaged_file_pathneeds_revision.npm run research:bundle -- approve --bundle <bundle-id>.npm run research:bundle -- publish --bundle <bundle-id> so staged JSON is promoted and a publication_event is written.npm run sync:research-planning after publish.npm run research:bundle -- smoke --bundle <bundle-id> --base-url <local-url> when a dev server is available.publishedpublication_event existsEvidence basis, finding inventory, and cited-source sectionspublished or rejected bundles.approved.