← ClaudeAtlas

researchzy-risk-of-biaslisted

Assess risk of bias per study, outcome and domain with a tool registered from its official file or page, each judgement backed by a quote checked against the saved full text; draw traffic-light plots; rate GRADE certainty per outcome. Use in a systematic review or meta-analysis.
Researchzy/Researchzy-academic-skills · ★ 0 · AI & Automation · score 72
Install: claude install-skill Researchzy/Researchzy-academic-skills
# Researchzy Risk of Bias This skill judges the risk of bias of each included study, result by result and domain by domain, and, when asked, rates the certainty of the evidence for each outcome (GRADE). The tool is registered for the review: the user gives its file, or you read its official page in this session, and `tool.json` records the version, source, date read and SHA-256. Every domain judgement carries a quote and its locator in the saved full text, or says that the paper gives no information and where you looked; a script confirms that each quote is in the text and that the overall judgement follows the tool's rule. Tables, plots and the summary of findings are written from the checked files. It bundles no tool: no signalling questions and no judgement algorithm, because tools get new versions and some licences forbid copying their text. It turns no judgement into a score or a quality band. Searching and screening belong to `researchzy-systematic-review` and pooling to `researchzy-meta-analysis` (both plugin `researchzy-evidence`, when installed); auditing a manuscript against a reporting checklist belongs to `researchzy-paper-review` (plugin `researchzy-manuscript`, when installed). ## Requirements Python 3.10 or newer; resolve `SKILL_DIR` to the directory containing this `SKILL.md`. `rob_check.py` and `sof_check.py` need only the standard library. `rob_plot.py` needs matplotlib: without it no plot is drawn and the run says so; ask before installing it. No accoun