Research Refine Pipeline

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

This chains together method refinement and experiment planning into a single pipeline, so you don't have to manually connect research-refine and experiment-plan yourself. It runs the full loop: takes a vague research direction, refines it through iterative reviews until you have a stable thesis, then builds a claim-driven experiment roadmap tied to what the paper actually needs to prove. The workflow enforces a gate between phases so you're not planning experiments on top of an unstable method. Outputs land in refine-logs as versioned markdown files, and you get a pipeline summary that tells you exactly what to run next. Good for when you want the complete package in one shot rather than toggling between refinement and validation planning.

skills.sh installs: 533
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 research-refine-pipeline --agent claude-code

Installs into .claude/skills of the current project.

Files

Research Refine Pipeline: End-to-End Method and Experiment Planning

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.

Refine and concretize: $ARGUMENTS