Result To Claim

wanshuiyin/auto-claude-code-research-in-sleep

This is the gate you run after experiments finish but before you write the paper. It collects results from W&B, experiment logs, or whatever you have, then sends them to Codex with the original claim you were testing. Codex judges whether the data actually supports what you want to say, flags missing evidence, and suggests whether you need to pivot, run more experiments, or just revise the claim. The pre-check is smart: it mechanically verifies that cited numbers exist in their source files before spending a Codex call, catching hallucinated evidence early. Routing is automatic based on the verdict. Use this when results are in and you need an honest second opinion before committing to claims.

skills.sh installs: 537
GitHub stars: 18.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add wanshuiyin/auto-claude-code-research-in-sleep --skill result-to-claim --agent claude-code

Installs into .claude/skills of the current project.

Files

Result-to-Claim Gate

Before each reviewer slot, follow Selectable Reviewers: resolve invocation → .aris/reviewer.json → Claude Code default (Codex MCP) with reviewer_route.py through the canonical helper resolver, preserving this slot's tier, fresh contexts, source/image inputs and deterministic gates. Codex examples/pins below apply only to backend=codex; other routes use their own first-call/reply contract. Save the route and actual returned identity; unavailable selections are REVIEW_UNAVAILABLE. Classify acceptance from actual families/completion/scope. Forward invocation reviewer/model/effort overrides to reviewer-bearing children. Generation workers are unchanged.

SKILL.md on GitHub

Details

Repository: github.com/wanshuiyin/auto-claude-code-research-in-sleep
Categories: Documentation, AI & Agent Building, Data Science & ML
First seen: Jun 3, 2026

Result To Claim

Install

Files

Result-to-Claim Gate

Details

More from wanshuiyin/auto-claude-code-research-in-sleep

Recommended