Busabase
Busabase is an open-source (MIT) database and workspace for AI agents — data, docs, skills, and apps in one place. Run it locally, self-host it inside your network, or use Busabase Cloud, and turn age
翻译准备中
尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证正在加载完整 README。 以下为已保存的摘录。
Unofficial geometric coverage clock for catalogued Rassvet / «Рассвет» objects over a point in Ukraine. Not affiliated with Bureau 1440 or any government. https://rassvet-over-ukraine.vercel.app is the current live deploy. Vercel project lizard-logic/rassvet-over-ukraine builds production from GitHub main. https://rassvet-over-ukraine.grok.me/ is an unsynced archive only: a one-shot Grok Build publish that cannot sync with this GitHub repo. It is not the self-managed live deploy.查看完整 README ↗
Unofficial geometric coverage clock for catalogued Rassvet / «Рассвет» objects over a point in Ukraine. Not affiliated with Bureau 1440 or any government.
https://rassvet-over-ukraine.vercel.app is the current live deploy. Vercel project lizard-logic/rassvet-over-ukraine builds production from GitHub main.
https://rassvet-over-ukraine.grok.me/ is an unsynced archive only: a one-shot Grok Build publish that cannot sync with this GitHub repo. It is not the self-managed live deploy. The Grok Build sandbox that produced the first commit (.grok skills, platform auth, and the Created with Grok banner) is not part of this repository. Agent notes are in AGENTS.md.
At a ground point P and minimum elevation E, how many minutes of geometric line-of-sight the current public catalog produces today, when the next window is, and which objects are still climbing versus near operational altitude (raised ≥ 450 km).
The default set is production RASSVET-3 at or above 450 km. Experimental Rassvet-1 and Rassvet-2 are separate choices. Below 450 km, a recent altitude history steeper than 0.8 km/day is climbing. A flat or falling rate is not climbing. When that rate is unknown, the status is rate pending and is not counted as climbing.
Visible is never presented as linked, encrypted, military-ready, or 24-hour service.
Follow docs/new-launch-group.md. Edit src/config/objects.json, then run npm run verify:catalog. The live Vercel site rebuilds from main.
Do not hardcode on-orbit counts in the UI. Counts are derived from the catalog plus status rules.
| Key | Meaning |
|---|---|
raisedAltitudeKm | 450 — Raised population gate. |
reentriesUrl | GitHub fallback for SATCAT decay records. Blob data/reentries.json is read first. RASSVET-3 4 (68363) decayed 2026-06-06. |
staleHours | 72 — mark and still propagate |
stepSeconds | 30 — elevation sampling |
minPassSeconds | 60 — ignore shorter passes |
earthRadiusKm | 6371 — spherical small-circle for the map footprint rings and the pass pre-filter. Elevation uses satellite.js look angles with a WGS-84 observer. Altitude is the Brouwer mean semi-major axis from satellite.js SGP4 init minus the WGS-72 equatorial radius 6378.135 km (μ = 398600.8 km³/s²). |
muKm3s2 | 398600.4418 — not used for altitude. |
data/gp-latest.json, then the GitHub data branch URL, then src/data/catalog-seed.json (snapshot dated in-file).RassvetOverUkraine/0.1 (unofficial geometric coverage tracker).npm run record:altitudes is the Blob writer (node scripts/record-altitude-history.mjs --blob). It needs BLOB_READ_WRITE_TOKEN and no other environment variable. It reads the prior history and reentries from the public Blob pathnames, merges, and writes them back. data/gp-latest.json is written only when every GP query succeeded. A CelesTrak fetch that still fails after retries exits 1 and does not update the files. A feed with no new epochs exits 0. Objects named RASSVET that are not in objects.json are a warning, and the count is printed as unlisted_rassvet. Run npm run seed:blob once before the first record so the log is the current data branch, not an empty file.
.github/workflows/altitude-history.yml still runs at 05:17 and 17:17 UTC, and when someone starts it by hand. It writes the same three files on the data branch and does not use Blob. Each run fetches the Rassvet GP JSON from CelesTrak once, appends {norad, name, epoch, alt_km} to altitude-history.json, and drops epochs older than 60 days. Those files are not on main. The run asks CelesTrak for the RASSVET name query and for each experimental launch query already listed in celestrak.queries. Rows are kept only for NORADs in the catalog, under the display name. It asks SATCAT only about a tracked object that is missing from those responses or whose epoch is more than 72 hours behind the newest one. A failed GP fetch retries, then fails the job. A feed with no new epochs succeeds and leaves the file as it was.
The site reads Blob first (data/altitude-history.json, data/reentries.json, data/gp-latest.json), then the raw data branch URLs, on the same ten-minute cache. A missing history file, a window that is too short, or a newest epoch more than three days behind the newest epoch in the catalog leaves that rate unknown. Under 450 km the status is then rate pending, which is not counted as climbing, unless the object is reentering: no decay date, that same stale epoch, and a last altitude under 250 km or a decaying rate. reentries.json holds SATCAT-confirmed decays. A later run drops a record when SATCAT no longer reports that decay, or when a GP epoch falls on a later UTC day than the decay date.
Per-object passes in the table. Today minutes and city chips union overlapping intervals so overlapping satellites do not double-count.
Language files: src/config/i18n/{en,uk,ru}.json plus Method copy in src/lib/i18n/method.ts. Review the homepage sentence with a human before quoting the numbers in print.
.cursor/environment.json points install and start at pstack setup. install runs bash .cursor/setup-cloud-agent.sh after checkout, and only during a snapshot build of the default branch. start is an inline command Cursor launches on every boot, including before that checkout has created the script. It waits until .cursor/setup-cloud-agent.sh and .cursor/pstack/pstack-models.mdc are both non-empty (bound PSTACK_CHECKOUT_WAIT_SECONDS, default 60), then runs the script. Each attempt appends timestamped lines to ~/.cursor/pstack-setup.log.
The tracked model map is .cursor/pstack/pstack-models.mdc. It stays out of .cursor/rules/ so a local Cursor window does not apply the committed copy and the home copy together. The script copies it to ~/.cursor/rules/pstack-models.mdc, which is the always-applied file pstack reads. Budget is large: code roles use grok-4.7-xhigh-fast; judgment, prose, and the hardest tasks use claude-opus-5-5-xhigh; arena, architect, and interrogate panels list both. If no pstack install is present, the script sparse-clones https://github.com/cursor/plugins and copies pstack/ to ~/.cursor/plugins/local/pstack.