baseline-review

스타0

포크0

업데이트2026년 6월 4일 14:46

Use when initializing baseline coverage for one bounded research scope unit by staging source, artifact, finding, and claim records into a candidate bundle.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

출처

tylerburleigh

tylerburleigh/lit-review-studio

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Baseline Review

Create a candidate bundle for exactly one taxonomy scope unit. Keep public claims traceable to staged findings and sources.

Read First

lit-review-studio.config.json and the active LIT_REVIEW_STUDIO_DOMAIN value.
domain-packs/<domain-id>/domain.json
domain-packs/<domain-id>/taxonomy.v1.json
domain-packs/<domain-id>/evidence-ladder.v1.json
domain-packs/<domain-id>/extraction-schema.v1.json
domain-packs/<domain-id>/appraisal-lanes.v1.json
domain-packs/<domain-id>/skills/baseline-review.md
research/state/coverage-status.v1.json
research/backlog/priority-queue.v1.json
Existing records under data/sources, data/artifacts, data/findings, data/claims, and data/candidate-bundles.

Workflow

Resolve the active domain and one target taxonomy node. If more than one scope unit is requested, split the work into separate bundles.
Inspect existing public records and active bundles for the same taxonomy node so the bundle does not duplicate live coverage.
Apply the domain adapter before drafting. Extract every required field from the domain extraction schema before writing a finding or claim.
Stage records under data/staged-records/<bundle-id>/. A baseline bundle normally includes one source, one artifact when applicable, one finding, and one claim.
Write data/candidate-bundles/<bundle-id>.json with intake_mode: "baseline_review", scoped taxonomy IDs, proposed changes, staged paths, required appraisal lanes, and source IDs/URLs.
Write a research session under research/sessions/ that records the scope, source screening, staged output, and next actions.
Run validation:

npm run validate
LIT_REVIEW_STUDIO_DOMAIN=<domain-id> npm run research:bundle -- validate --bundle <bundle-id>
npm run sync:research-planning

Boundaries

Do not publish, approve, or bypass evidence appraisal.
Do not edit live public records directly; only stage proposed records and create a bundle.
Do not turn activity or source discovery into a public claim unless a source-backed finding supports it.
Keep limitations near the claim, especially when the domain adapter marks them as required.

Expected Outputs

Staged JSON records under data/staged-records/<bundle-id>/.
One candidate bundle under data/candidate-bundles/.
One research session under research/sessions/.
Passing npm run validate and bundle validation for the active domain.

이 저장소의 다른 Skills

같은 저장소

evidence-synthesis

tylerburleigh/lit-review-studio

Use when producing a cross-record synthesis report for one domain or bounded set of taxonomy scope units after baseline review or review update records have been published.

2026-06-040

editorial-decision

tylerburleigh/lit-review-studio

Use when deciding whether an appraised candidate bundle should receive comments, changes requested, rejection, approval, or publication.

2026-06-040

evidence-appraisal

tylerburleigh/lit-review-studio

Use when verifying one appraisal lane for one candidate bundle revision before editorial approval or publication.

2026-06-040

review-update

tylerburleigh/lit-review-studio

Use when checking an already scoped research unit for material changes since the latest baseline-review pass, publication, or review-update pass.

2026-06-040

literature-review-drafting

tylerburleigh/lit-review-studio

Use when drafting or revising paper-facing literature-review prose from already-collected research records, especially background sections that need inline source citations for substantive claims.

2026-06-040

review-triage

tylerburleigh/lit-review-studio

Use when deciding what needs attention now in the active lit-review-studio domain, including coverage gaps, bundle readiness, source-access issues, source-summary gaps, and thin claim support.

2026-06-040

Baseline Review

Create a candidate bundle for exactly one taxonomy scope unit. Keep public claims traceable to staged findings and sources.

Read First

lit-review-studio.config.json and the active LIT_REVIEW_STUDIO_DOMAIN value.

domain-packs/<domain-id>/domain.json

domain-packs/<domain-id>/taxonomy.v1.json

domain-packs/<domain-id>/evidence-ladder.v1.json

domain-packs/<domain-id>/extraction-schema.v1.json

domain-packs/<domain-id>/appraisal-lanes.v1.json

domain-packs/<domain-id>/skills/baseline-review.md

research/state/coverage-status.v1.json

research/backlog/priority-queue.v1.json

Existing records under data/sources, data/artifacts, data/findings, data/claims, and data/candidate-bundles.

Workflow

Resolve the active domain and one target taxonomy node. If more than one scope unit is requested, split the work into separate bundles.

Inspect existing public records and active bundles for the same taxonomy node so the bundle does not duplicate live coverage.

Apply the domain adapter before drafting. Extract every required field from the domain extraction schema before writing a finding or claim.

Stage records under data/staged-records/<bundle-id>/. A baseline bundle normally includes one source, one artifact when applicable, one finding, and one claim.

Write data/candidate-bundles/<bundle-id>.json with intake_mode: "baseline_review", scoped taxonomy IDs, proposed changes, staged paths, required appraisal lanes, and source IDs/URLs.

Write a research session under research/sessions/ that records the scope, source screening, staged output, and next actions.

Run validation:

npm run validate LIT_REVIEW_STUDIO_DOMAIN=<domain-id> npm run research:bundle -- validate --bundle <bundle-id> npm run sync:research-planning

Boundaries

Do not publish, approve, or bypass evidence appraisal.

Do not edit live public records directly; only stage proposed records and create a bundle.

Do not turn activity or source discovery into a public claim unless a source-backed finding supports it.

Keep limitations near the claim, especially when the domain adapter marks them as required.

Expected Outputs

Staged JSON records under data/staged-records/<bundle-id>/.

One candidate bundle under data/candidate-bundles/.

One research session under research/sessions/.

Passing npm run validate and bundle validation for the active domain.

name	baseline-review
description	Use when initializing baseline coverage for one bounded research scope unit by staging source, artifact, finding, and claim records into a candidate bundle.