Skip to content
Get packs

Proof Orchestrator

Runs multi-session proof projects with local-first attempts, manual GPT Pro handoff packages, and optional DeepSeek review.

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

What it does

Once installed, the agent organizes a mathematical proof as a numbered, resumable run folder. It first tries the proof locally, checks it, and tidies the result. If it gets stuck, it writes a self-contained prompt you can paste into GPT Pro in your browser, and it keeps track of what has been proved, what is still conjectural, and what sources were used, so a later session can pick up where the last one stopped. It does not operate a browser or spend API credit by default.

Great for

  • Mathematicians or students working on a long proof that spans several sessions
  • Researchers who want a clean audit trail of which claims were proved locally and which came from an outside model
  • People who already use GPT Pro or a similar chat tool and want a structured handoff prompt

Skip it if

  • Quick, one-off proof drafting; the SKILL.md routes that to a separate /proof-writer skill
  • Final verification before paper submission; this skill points to /proof-checker for that gate, so relying on it alone could miss errors
  • Anyone expecting automatic verification: the output is a model's reasoning and still needs expert review

Try asking

  • “Continue the proof from last week's run directory and identify the next narrow obligation”
  • “Prepare a GPT Pro handoff package for the lemma that the local attempt could not close”
  • “Get an independent DeepSeek second opinion on the current proof run as extra evidence”
  • “Audit the notation in this proof and report the symbol scorecard”
Step-by-step guide: how to use Proof OrchestratorInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the MIT licence and an attribution note.

    proof-orchestrator.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip proof-orchestrator.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip proof-orchestrator.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

  • Give the agent the exact theorem statement and any source papers up front, since the skill relies on clean source snapshots
  • Use the manual handoff by default; only ask the agent to call GPT Pro or DeepSeek when you are comfortable with that data leaving your machine
  • Review gpt-pro-output.md and audit.md yourself before trusting any claim marked proved

What's inside

  • NOTICE.md
  • references/audit-output-contract.md
  • references/deepseek-routing.md
  • references/dispatch-prompts.md
  • references/notation-audit.md
  • references/proof-audit-rubric.md
  • references/stress-tests.md
  • 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.