Twelve instruction sets you paste into an AI chat workspace before you hand it a folder of documents. Each module fixes the role, the variables you fill in, and the table the output must come back in — so a run can be checked line by line instead of taken on trust.
12 modules · English and Chinese editions · 9 files · DOCX, PDF, TXT, CSV
Checkout is opening shortly — the buy link is added before public launch.
Not a course, and not a gallery of templates. It is a set of operating instructions for the half hour before batch work starts: what the assistant may touch, what it must not decide, and what it must hand back.
Every module locks three things at once — the role boundary, the variables you fill in (folder, naming rule, template fields, edit tolerance), and the exact table the output must arrive in. A short self-check line sits under the table, so you can tell whether a run actually followed the instruction or quietly improvised around it.
Lists what is in a folder, flags suspected duplicates, applies your naming rule, and returns a rename map before anything is moved.
Plans the conversions and checks what survived — heading levels, tables, images, comments — then logs what did not, per file.
Pulls conclusions, evidence, to-dos and risks, and pins each one to a section or page number instead of paraphrasing freely.
Fills a fixed summary table plus a commitments table, keeping stated commitments apart from intentions still to be confirmed.
Typos, terminology, punctuation, numbering, units and dates — each with location, original fragment, suggested replacement and severity.
Maps source content onto a target template without adding material, inventing examples, or shifting the position the source took.
Rebuilds tables into a fixed column set, records merged cells, footnotes and missing values, and marks what could not be recovered.
Fills form fields from documents and marks gaps as gaps — the module is written so that a blank never gets filled by inference.
Compares figures, names, dates and defined terms across a document set, and lists each conflict with both of its sources.
Builds a navigable index with section, page and topic, so a hundred-file set stops being a folder and starts being a catalogue.
Runs a checklist against a finished batch and reports pass or fail per item, including the items that were simply not attempted.
Fixes naming, folder layout and versioning, and writes the handover note that says what changed and what is still open.
Choose the one that matches the job. The modules are deliberately narrow; running two of them on one batch is usually cleaner than bending one.
Replace the bracketed fields — folder, naming rule, template, edit scope, tolerance. These are the only places you are meant to edit.
Read the output table and the self-check line before accepting the batch. If a row is missing, the run did not follow the instruction.
Nine files, about 550 KB, delivered as a download link after checkout.
Two editable DOCX (English and Chinese). Two fixed-layout PDF (25 pages Chinese, 29 pages English). Two plain-text instruction bodies you can paste without opening Word. A bilingual glossary in CSV, eighteen terms. A QA report that states what was measured and what it does not cover. A manifest listing every file in the package.
This pack contains instructions for an AI assistant, not AI output. Results depend on the model you use, the documents you feed it, and your own review. Nothing here guarantees a particular outcome, a time saving, or any compliance status.
It is an independent publication. It is not affiliated with, endorsed by, or sponsored by any AI platform, word processor or standards body; product and file-format names appear only to describe compatibility.
Nothing in the pack is legal, tax or certification advice. Where a module touches personal data, commercial secrets or regulated records, redaction, retention and lawful basis remain your responsibility.
US$29one-time · instant download
Prices in US dollars. No subscription, no account with us.