manuwright
Medical manuscript workflow with the ManuWright engine. Use when planning, drafting, revising or checking a paper in a folder that has project.json, drafts/draft_plan.md or knowledge/evidence.md, or when the user mentions ManuWright, draft plan, analysis plan, evidence.md, EVID citations, reviewer response or submission build.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1.6k
- Forks
- 491
01 前言 Front Matter
- name
- manuwright
- description
- Medical manuscript workflow with the ManuWright engine. Use when planning, drafting, revising or checking a paper in a folder that has project.json, drafts/draft_plan.md or knowledge/evidence.md, or when the user mentions ManuWright, draft plan, analysis plan, evidence.md, EVID citations, reviewer response or submission build.
03 SKILL.md 正文
ManuWright workflow
- Load the rules for the current step:
manuwright rules <keyword>(for exampleCitation Integrity,Draft Plan,Verification Gates,Phase 4). Runmanuwright rulesonce for the full workflow when the phase is unclear. - Phase order: setup and evidence, then analysis plan (approved), analysis, draft plan (approved), drafting (Methods, Results, Introduction, Discussion, Conclusion, Abstract, Title), style polish, QC (3+ rounds), finalize, revision.
- Plan first. If
drafts/draft_plan.mdordata/analysis_plan.mdis missing, incomplete or not approved by the author, stop and help complete it. Show the plan to the author. When they approve it explicitly in chat ("승인", "approve"), record it withmanuwright approve <plan> --kind analysis|draft --approved-by "<author>" --quote "<their exact words>"(ticks the box, notes who/when/what, writes the hashed receipt). Never approve on your own judgement or from a general "continue". - Grounding. Cite only
[EVID:id]keys inknowledge/evidence.mdand use only numbers inresults/*.csv. Register new sources inevidence.md(verified PMID/DOI) before citing them:manuwright search "<query>". - Analysis code runs in the paper's own environment:
manuwright env(once; uv-managed Python with the packages indata/requirements.txt), thenmanuwright run data/py/<script>.py. Do not install packages into, or debug, the system/Homebrew/pyenv Python; add packages todata/requirements.txtand rerunmanuwright env. Cite the versions indata/environment.lock.txtin Methods. - Verify before saying a step is done (see the
manuwright-verifyskill). - Academic writing: before drafting or rewriting a section, run
manuwright style card <section>and write to that card (moves, rules, phrasebank, register of the model paragraphs, the author's learned corpus style). Never reuse the model paragraphs' facts, numbers or citations. Fix the academic-prose findings reported after each edit; instrictmode a write with high-severity findings is blocked until rewritten. After a style rewrite of existing text,manuwright style preserve <before> <after>must print OK. - Evidence and revision safeguards: record
Claim Strengthfor each evidence.md entry and runmanuwright claim-strength drafts;manuwright search audit(metadata match, retractions) before submission; letter-blind re-review (manuwright blind-review packet/check) before recording response_alignment;manuwright style editsproposes rules from the author's edits, which only the author approves. When the author asks in chat, run the tool for them: "내 문체 배워줘" / "learn my style" ->manuwright style learn <folder>(3+ papers); "내가 고친 거 배워줘" ->manuwright style edits --git <rev>(or two files), show the proposals, and after they approve a rule in chat runmanuwright style edits --approve "<rule>" --approved-by "<author>" --quote "<their words>" --apply; "과장 표현 검사해줘" ->manuwright claim-strength drafts; "참고문헌 다시 확인해줘" ->manuwright search audit; "블라인드 재검토" ->manuwright blind-review packet .... - Guides live in the engine, not in the paper folder: read any
docs/<name>.mdthe rules cite withmanuwright guide <name>(several at once allowed;manuwright guidelists them). Everymanuwright rulesoutput ends with the guides it cites; read the ones for the current step in full before doing it.
Journal fit and suggestions
"Does it fit Spine J?" / "which journal?": build a PubMed query from the population, intervention and outcome, then manuwright journal fit "<NLM abbreviation>" --query "..." or manuwright journal suggest --query "..." [--review]. With no journal profile, create one (manuwright journal profile "<journal>" --url <official page>), read the official Aims & Scope / Guide for Authors yourself and fill it; mark it verified only after the author confirms. Never state an impact factor, acceptance rate or policy that is not in the profile (docs/critical_review_protocol.md section 6).
Personal library (reused across papers)
manuwright library shows where it is (~/.manuwright/library/, or $MANUWRIGHT_HOME/library). New papers get copies from it at manuwright init; Word styles are chosen per paper in manuwright target.
"Register my writing style" (the author added papers with manuwright library writing add <pdf|md> --kind own|landmark|target_journal):
- Read the engine's
Style/style_guide.md(folder next tomanuwright rules --path) and follow its Extraction Framework and copyright policy: patterns, metrics and short phrases, never copied paragraphs. - For each source in
library/writing/PDF/<kind>/without a matchinglibrary/writing/<kind>/<same basename>.md, write that markdown anchor. - Measure instead of guessing: run
manuwright style learn(it reads the library PDFs, or pass files) so every section card carries the measured style and model paragraphs, andmanuwright style extract <text or md files>for sentence length and citation density in the anchors. - Update
library/writing/terminology.md(preferred and forbidden terms with context) and writelibrary/writing/style_spec.mdfrom the engine'sdocs/style_spec_template.md(manuwright guide style_spec_template), bound to the author's own papers. - Show the author a short summary of the voice you found and the files you wrote; do not change any paper folder.
"Fill my team profile" (from a CV, author list or earlier title page the author gives you): edit library/profile/authors.md (manuwright library profile creates it from the template). Copy names, degrees, affiliations, ORCID, emails and grant numbers exactly as given; leave [...] placeholders for anything not given; never guess an ORCID or grant number. Ask the author to check it. Papers created later copy it to profile/authors.md; an existing paper keeps its own copy.
Keep what the author teaches across papers (memory). When the author asks you to remember something ("기억해", "앞으로는 X 대신 Y", "공동저자 추가해줘", "소속이 바뀌었어", "remember", "from now on"), save it in the personal library, additions only, and say in one line what you saved where:
- a wording rule:
manuwright library term --prefer "X" --avoid "Y" [--context "..."](library and this paper; a CONFLICT means the library already prefers another word: ask); - one author, affiliation, ORCID, email or grant: edit only that person's lines in
library/profile/authors.mdand in this paper'sprofile/authors.mdif it exists; show the changed lines first; never delete other people or guess numbers; - any other lesson (journal requirements, reviewer preferences, method choices):
manuwright library note "<one sentence>" --topic journal|reviewer|writing|method|other; notes appear at every session start in paper folders. "라이브러리 최신으로 받아줘" =manuwright library sync --pull; "이 논문 것을 라이브러리에 올려줘" =manuwright library sync --push; "차이 보여줘" =manuwright library sync. Sync only adds; a "differs" line (style spec, team profile) is merged by you on request, line by line, with the author's confirmation. Rules the author approves withstyle edits --applygo to the library automatically, and a session in a paper folder pulls the library's additions automatically (manuwright config set library-sync auto|ask|off).
04 同仓库更多 Skills
worktable-review-with-annotations
Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.
worktable-find-and-synthesize-context
Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.
worktable-create-or-update-docs
Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.
worktable-create-or-manage-records
Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.
worktable-create-or-manage-interactive-html
Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.
worktable-collaborate-threads
Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.