← ClaudeAtlas

paper-review-litelisted

Run a pre-submission manuscript audit covering argument, numerics, references, writing, figures, methods, preregistration, and replication readiness.
scdenney/open-science-skills · ★ 39 · AI & Automation · score 73
Install: claude install-skill scdenney/open-science-skills
# Paper Pre-Submission Review (Lite) ## Heritage and scope An 11-subagent pre-submission review that runs inside a Codex session with no extra review tool to install. Nine parallel dimension agents find issues, two cross-checkers verify each finding against the manuscript before it reaches the report, and the lead synthesizes. This is the Codex-native counterpart to [`presubmit`](https://github.com/scdenney/presubmit), itself descended from the [reviewer2](https://github.com/isitcredible/reviewer2) adversarial peer-review pipeline, and it inherits that lineage's two commitments: a critical-reviewer posture, and a verification cascade that drops any finding not pinned to a quoted passage. It is the fast in-flow check, not the full pipeline — see the last section for when to reach for `presubmit` instead. ## Delegation gate Run the subagent workflow only when the user explicitly invokes `$paper-review-lite` or asks for parallel/subagent review. If the skill loads implicitly, either perform the dimensions sequentially in the lead context or ask before spawning agents. ## Instructions Nine agents each audit a different dimension; two cross-check agents then audit those Phase 2 findings for hallucinations, false positives, and missed issues. The deliverable is a severity-ranked Pre-Submit Report with a journal-readiness checklist. **Critical-Reviewer posture (required for every sub-agent).** Brutally honest on the work, fair to the people. Attack the *argument or the data*,