Dashboard

candidate-search

Enumerate Nielsen class window 184

One discovery run plus two deterministic replays before dashboard credit.

Statusavailable
Priority96
Runtime25-90 minutes
Remaining9 units

Safety task card

你和你的 AI 会算什么

What you will compute

Enumerate a public Nielsen-class search window and return candidate branch-cycle hashes, not a proof claim.

Why it matters

This narrows which branch-cycle regions deserve deterministic replay for possible M23 inverse Galois candidates.

Public inputs

  • branchCyclePattern=2A-4A-23A
  • window=184
  • seed=230184
  • maxDepth=12

Data sent to project

  • candidate hashes
  • checker status
  • logs hash
  • runtime
  • environment digest

Data never requested

  • AI API keys or provider credentials
  • SSH keys, browser cookies, tokens, or passwords
  • Private files, contact lists, wallet data, or local browser history
  • Network access to anything except the project API and the user's chosen AI provider

Prohibited actions

  • Execute arbitrary code supplied inside a work-unit response
  • Upload raw API keys, prompts containing secrets, private local files, or unrelated telemetry
  • Let an AI agent browse the local machine or external websites beyond the declared task
  • Mark AI output as verified without deterministic checker acceptance

AI scope

AI may propose branch-cycle search heuristics from the public task card only; it must not inspect private files or credentials.

Submissions

Recent evidence for this task

No submissions yet. The first accepted result will appear here after upload.