为什么值得关注

它通过保留每个快照实现数据溯源和不可变性，通过字节匹配检查强制发布完整性，并提供安全的双检出同步机制，且无需第三方 Python 依赖。

适合谁使用

寻找 AI-for-science 项目资金或计算资源的研究人员
管理科学计算资源配置的资助管理人员
需要透明且可验证资源追踪的开放科学倡导者
构建依赖精心策划、版本化机会数据的开发者

典型使用场景

发现当前开放的科研计算资助和云 API 信用
通过不可变快照审计历史资助机会
在主站和镜像之间同步相同数据集并确保完整性
在发布新快照前验证数据完整性和来源

项目优势

不可变性：每个记录状态都通过生命周期字段和完整历史快照保留
零外部依赖：构建和验证仅使用标准库 Python 3.10+
严格测试：全面的测试套件验证模式、边界、生命周期不变量和同步
安全同步：同步助手确保在不同检出间逐字节一致分布，无需删除或意外编辑

使用前须知

快照日期为 2026-07-29，可能无法反映当前资助的实际可用性
发布需要手动操作和本地命令审核，没有自动化部署
数据集较小：初始快照仅包含 48 条公共记录

README 快速开始

Scientific Resources

A static, official-source tracker for scientific compute allocations, cloud and API credits, and AI-for-science or open-science grants.

Primary:
Synchronized mirror:
Source repository: Earth-Space-AI/scientific-resources

The inspected primary route normalizes its canonical and social URL to the slashless https://earth-space-ai.org/scientific-resources. The mirror intentionally remains self-canonical at https://huangzesen.github.io/scientific-resources/ and links to the primary as its alternate.

Snapshot scope

The canonical public dataset is a reviewed snapshot dated 2026-07-29 (America/Los_Angeles), with collection verification timestamp 2026-07-30T01:16:56Z. The configured Pacific date did not roll over. The dataset contains 48 public records: 32 actionable open, 1 actionable upcoming, and 15 archived. These are preserved snapshot facts, not a claim that the same statuses remain current today.

A specific modeled cycle with an officially verified fixed applicant deadline on or before page_date + 15 calendar days is moved out of the actionable view. The boundary is inclusive and group-neutral. Policy archival is not a claim that the sponsor closed intake: the record preserves its official deadline and sources, removes its apply control, and shows a public retirement_reason. A continuously open program with recurring cutoffs advances to the next verified cutoff outside the fence rather than closing the whole program; a multi-element umbrella has no single fixed deadline.

data/opportunities.json is the sole canonical current opportunity data file. Every stable ID remains present after publication; retirement is represented with lifecycle fields, not by deleting the record. Each current record carries first_seen, last_verified, retired_at, retirement_reason, superseded_by, and reactivated_at. The exact original 32-record dataset is preserved as the first immutable history snapshot at data/history/snapshots/2026-07-28-a5af1ed92d34/.

Build

Requirements: Python 3.10 or newer; no third-party Python packages.

python3 src/generate.py

The production build writes the current data, provenance, and immutable history files to each host profile:

dist/primary/index.html
dist/primary/public_opportunities.json
dist/primary/provenance.json
dist/primary/snapshots/in

项目描述

Public, reproducible tracker for scientific compute, cloud/API credit, and research funding opportunities.

scientific-resources