Skip to content
arcacomputerPublic

About

Arca OSS public engineering index: maintained projects, upstream support, and evidence-backed contribution ledgers.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Arca OSS

Source for oss.arcabot.ai, Arca's public engineering index.

The site has two primary destinations: Software we make (#projects) for Arca's licensed software, and Projects we contribute to (#contributions) for external OSS work, with current and past contributions beneath it. Search and category/outcome filters help visitors explore each listing. Public records holds the GitHub tracking sources, full public repository inventory, and detailed evidence. Main GitHub and site-source links point to arcacomputer/oss.

The software directory and external contributions are separate. Software we make includes tools, Ardea Knowledge Steward, and OpenChina. Headlong Agent Findings, the OpenClaw contribution ledger, the TUI field fix, and this index appear as compact resources in Public records; their underlying project evidence remains available. Contribution records distinguish open, merged, closed, superseded, reviewed, and co-authored work with precise attribution. Public personal repositories are supporting records, not automatically software made by Arca.

GitHub tracking scope

Authored upstream PR tracking covers public external repositories for felirami and arcabotai, with PRs created on or after February 12, 2026. Repositories owned by either user or arcacomputer are excluded from upstream contributions. Earlier receipts, including the 2023 SlimeVR contribution, are outside this scope.

The repository inventory covers all currently public repositories owned by felirami, arcabotai, and the arcacomputer organization, regardless of creation date. It labels forks, archived repositories, and detected licenses separately; public visibility alone does not establish an open-source license or Arca authorship. Reviews and coauthor credits are selected, verified receipts, not a complete discovery of those activities.

Architecture

  • Astro 7 static output
  • React renders the page at build time; no React hydration or application runtime
  • Small progressive-enhancement script for project search, category filters, activity filters, and mobile navigation
  • Self-hosted DM Sans variable font with its OFL license
  • Cloudflare Workers Static Assets
  • HTML-only Cache-Control: no-transform prevents Cloudflare from auto-injecting analytics JavaScript
  • No application server, database, authentication, or runtime rendering

Evidence refresh

The previous Next.js site fetched the OpenClaw contribution ledger with one-hour ISR. The Astro site uses committed, reviewable snapshots:

  • npm run refresh:activity refreshes public/activity.json and public/github-repositories.json every six hours through the activity workflow
  • public/openclaw-prs.json refreshes hourly
  • each workflow validates claims and commits the public snapshots, including successful checkedAt freshness updates
  • every committed refresh triggers the Cloudflare deployment workflow; the live site advances only after checks, the dependency audit, upload, and promotion succeed

Hermes contribution receipts distinguish our directly authored merge #97871 from maintainer-incorporated credit #76400. These two contributions are verified in the v0.21.6 source tag; that does not establish inclusion in a packaged Desktop binary. Later incorporation receipts do not inherit that release claim.

Contribution outcomes (additive schema v1)

oss.json is the single curated source for mergedUpstreamCredits and manuallyIntegratedContributions. The original PR's state and evidence.mergeCommit remain GitHub disposition, not a contribution-outcome shortcut:

  • upstream_credit events are merged replacement PR receipts, with the credited actor separate from evidence.prAuthor. landedCommits records actual landed author/co-author evidence; legacy implementationCommit fields may refer to pre-rebase branch commits and are retained as historical provenance.
  • upstream_manual_integration events use state: integrated, not merged. One origin contains all integration/backport commits and pinned branch tips; OpenSSL's master and openssl-4.0 copies count as one contribution.
  • Closed originals gain a separate contributionOutcome (incorporated, partially_incorporated, or manually_integrated) with linked receipts and limited attribution context. Incorporated originals are not grouped as unsuccessful closed-only proposals, nor added to authored-merge totals.
  • Refresh rebuilds outcomes only from freshly verified receipts, stripping previous derived outcomes first. A configured replacement URL, an open PR's synthetic merge SHA, or a closure comment alone cannot establish acceptance. Replacement verification requires structured merge metadata, exact curated origin/reuse text, landed author/co-author identity and ancestry to the merge. Manual verification requires retained authorship, an exact Merged-from trailer and ancestry to pinned branch tips.
  • scripts/contribution-claims.mjs, called by check:claims, reuses the refresh verifiers, compares public receipt payloads, recomputes outcomes and guards the original closed-unmerged disposition. Missing/invalid evidence fails the refresh/claim gate rather than publishing an acceptance claim. These historical integration checks do not prove current byte-for-byte retention or release inclusion.

Authored PR merges, distinct credited replacement PR merges and manual integrations are counted separately. Replacement URLs deduplicate credited counts; origin URLs deduplicate manual contributions. The eight audited origins currently correspond to eight credited replacement PRs plus one manual integration: ClickClack's one partially incorporated origin has two replacements, while OpenSSL has no replacement PR. The elizaOS additions are historical provenance only; diagnosis-only credit and merely superseded proposals do not become code-incorporation receipts.

A failed refresh leaves the last committed public snapshots intact. lib/data.ts provides curated fallback evidence within the tracking scope. lib/presentation.ts resolves historical transitions before showing current work, reconciles the OpenClaw ledger, and keeps authored merges separate from credited merges. lib/contributions.ts groups in-scope evidence by upstream project, including selected reviews and closed proposals. lib/catalog.ts provides plain-language project descriptions and lifecycle labels. Collection paginates GitHub results and fails explicitly on incomplete results; the activity snapshot has no 200-event cap. Snapshot check dates remain visible beside the counts.

Development

npm install
npm run dev

Verification

npm run check
npm run build
npm run check:claims
npm run cf:dry-run
npm run cf:dev
npm run smoke
npm audit

Public data

  • /oss.json
  • /activity.json
  • /github-repositories.json
  • /openclaw-prs.json
  • /llms.txt
  • /robots.txt
  • /sitemap.xml

Deployment boundary

Production is the arca-oss Cloudflare Worker serving oss.arcabot.ai. Its custom domain is already attached. wrangler.json defines no routes and disables workers.dev and preview URLs. Routine deployments upload and promote an exact Git commit without changing DNS or domain attachments.

License

MIT

About

Arca OSS public engineering index: maintained projects, upstream support, and evidence-backed contribution ledgers.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Used by

Contributors

Languages