Skip to content
Get packs

Full Research Pipeline

Runs an end-to-end autonomous research workflow from idea discovery through experiments and review to a paper draft.

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
1 file

What it does

This skill tells an AI coding agent to run a full research project in stages: surveying literature from a broad research direction, generating and filtering ideas, running experiments, having the work reviewed by a second model (via Codex MCP), and producing a narrative report that can feed into paper writing. Progress is saved to a run-state file so an interrupted run can be resumed rather than restarted. Several stages are delegated to other skills in the same repository, and the README-level settings control how much runs without human confirmation.

Great for

  • Researchers who want a long, mostly unattended exploratory run on a topic and will review the outputs afterwards
  • ML or science teams testing whether an automated pipeline can speed up idea-to-experiment cycles
  • Developers who already have Codex MCP and GPU or compute access set up

Skip it if

  • Anyone expecting a publication-ready paper without expert review: the skill itself says quality verdicts rest on its reviewer loop, and results still need human checking (reason: automated claims can be wrong or overstated)
  • Users without the required reviewer setup or compute, since several stages depend on external tools and experiment infrastructure (reason: the run will stall or fail partway)
  • Work with sensitive or proprietary data sent to external services (reason: the skill uses web search, web fetch and a second-model reviewer)

Try asking

  • “Run the full research pipeline on 'sparse attention methods for long-context language models' and stop at the narrative report”
  • “Resume run abc123 with /research-pipeline — resume abc123 after my machine restarted”
  • “Run the pipeline with human checkpoints turned on so I can review each review round before fixes are applied”
Step-by-step guide: how to use Full Research PipelineInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    research-pipeline.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip research-pipeline.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip research-pipeline.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Start with AUTO_PROCEED set to false or HUMAN_CHECKPOINT on if you are new, so you can inspect early choices before the pipeline commits compute
  • Set a specific VENUE only when you reach the paper-writing stage; without it, the pipeline defers writing rather than guessing
  • Review the generated NARRATIVE_REPORT.md and the reviewer outputs yourself before relying on any claimed result

What's inside

  • SKILL.md

Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.