Skip to content

2026-09-25 — Edit registry for rocky-hq org rename to erid-tech


2026-09-25 — Edit registry for rocky-hq org rename to erid-tech

Section titled “2026-09-25 — Edit registry for rocky-hq org rename to erid-tech”

Date: 2026-09-25 Status: ratified Verb: petrova_act_registry_edit Slug: rocky-hq MR-compliance: MR-7, MR-12

GitHub org rocky-hq was renamed to erid-tech (verified via gh api orgs/rocky-hq → 404, gh api orgs/erid-tech → 200, org display name still “ROCKY”, blog field https://erid.tech). Repo names under the org are unchanged (rocky, console, ralph, lore, hearth, algo, contracts), only the org segment of each URL moves — structurally identical to the chronicle-hq→chronicle-cat rename (docs/decisions/2026-09-25-registry-edit-chronicle-cat-rename.md) and to skyflow-hq’s own prior skyflow-me→skyflow-blue rename recorded in its registry notes.

The registry slug is kept as rocky-hq — internal identifier, not a live pointer; renaming it would cascade into state/rocky-hq.yaml, this domain file’s filename, cli/src/cairnet/fleet/enumerate.ts’s slug-keyed kit-home check, and every cross-reference in prior decision docs and audit sweeps, none of which need to change to fix the actual break (a stale url). Only registry.yaml > repos[rocky-hq].url is corrected.

This is not cosmetic. Confirmed via a direct GET /app/installations call (App JWT minted locally, read-only) that the PETROVA GitHub App installation followed the org rename automatically — installation login is now erid-tech (id 131909394) — but host/src/github-auth.ts’s auto-discovery resolves installations by extracting owner from registry.yaml’s url. With the stale rocky-hq url, every write verb against this slug was missing the discovery map and silently falling through to the PETROVA_APP_INSTALLATIONS/PETROVA_APP_INSTALLATION_ID env fallback — the exact failure mode skyflow-hq’s own notes already documented for its rename. The url fix alone restores correct resolution; auto-discovery is self-healing by design once the url is correct, no App reinstall or env change needed.

Also fixed in the same session, separately committed: two hardcoded ROCKY_REMOTE clone-source constants in cli/src/verbs/cairnet_onboard.ts and cli/src/cairnet/fleet/kit-run.ts — these do a live shallow-clone during CAIRNET onboarding and were 404ing for any non-rocky fleet member onboarding against the dead org URL. Not a registry concern, but the same root cause; both now point at erid-tech/rocky.

registry.yaml > repos[rocky-hq]:

  • url: https://github.com/rocky-hq/rocky → https://github.com/erid-tech/rocky
  • notes: appended rename record (installation-continuity verification, cross-reference to this doc).

docs/domains/rocky-hq.md regenerated via npm run domains:generate (host package) — only the repo URL line changed; parts 3–5 (author-owned, unspecified) untouched by the generator’s non-overwrite rule.

cli/src/verbs/cairnet_onboard.ts, cli/src/cairnet/fleet/kit-run.ts — ROCKY_REMOTE constant updated (separate concern, committed alongside).

  • Local checkout (~/code/workspace/rocky-hq) still has stale remotes (root + submodules) and .gitmodules pointing at the dead org — housekeeping only, no repo-side consequence, scoped as a later taskset.
  • console/package.json’s npm scope @rocky-hq/contracts is a separate namespace decision, not auto-renamed here.
  • CLAUDE.md’s Andy Weir/PHM naming-theme note (rocky-hq → “Rocky” character) remains accurate — the repo is still literally named rocky; only the org wrapper renamed. No change needed there.