De-duplicate, screen, and log decisions without losing your mind
A librarian-friendly, researcher-friendly pipeline for evidence synthesis. Import citations from multiple databases, de-duplicate, set screening rules, track decisions, and output counts plus audit logs for transparency and reproducibility.
Create a skill called "Systematic Review Dedup & Screening Pipeline". Intake questions: - Review question (PICO or equivalent; if not applicable: unspecified) - Databases searched and export formats available (RIS/BibTeX/CSV) - Screening criteria (inclusion/exclusion) - Screening mode: single or dual; who the screeners are (if unspecified, ask once then proceed) Workflow: 1) Validate imports (detect empty fields, encoding issues). 2) De-duplicate: - Exact matches - Probable matches flagged for human review - Produce a dedup log 3) Screening setup: - Build a screening form aligned to criteria - Build a reason-for-exclusion controlled vocabulary 4) Output: - Included list + excluded list + log - Counts summary suitable for methods reporting Rules: - Never delete anything without producing a log of what changed. - If a detail is unknown, write "unspecified" and continue.
Systematic reviews pull from multiple databases, producing duplicates and a heavy
screening burden. Reference managers only partially de-duplicate, leaving a
time-consuming manual tail. This recipe creates a structured pipeline with a
full audit trail.
File naming, versioning, and documentation that prevents chaos
Set up a research project the right way from day one: consistent folder structure, explicit naming conventions, versioning rules, and minimal documentation so "future you" (or collaborators) can understand the project without a tour.
Draft data management and required plans without missing key sections
Many grant processes require Data Management Plans (DMPs), and they're time-consuming but partially reusable. This recipe produces a structured draft with explicit placeholders for unknowns and a checklist to finalize institutional specifics.
Wikipedia-grade AI pattern removal
Comprehensive AI writing cleanup based on Wikipedia's WikiProject AI Cleanup guidelines. Catches 24+ distinct patterns including inflated symbolism, em dash overuse, rule of three, copula avoidance, and sycophantic tone.
Make invisible labor visible and ownable
Convert the invisible cognitive labor of being the "default parent" into a shared, ownable task system with clear accountability. One owner per domain — sees it, plans it, does it, confirms it.