Admin group v0.1.0 — Twenty CRM
Admin group — v0.1.0
Twenty CRM (server + worker + Redis) on the shared Postgres, one AWS Lightsail 8 GB
host. Per-group reporting infra present; curated Twenty views deferred
(metadata-driven dynamic schema) — use Twenty's native GraphQL/REST API. Caddy
auto-TLS; N2. Merged via PR #2.
Verified locally (Colima): 11/11 smoke (server migrated 24 tables, worker, PostgREST).
Captured gotchas (now reusable Postgres-init conventions): apps that self-create
their DB need CREATEDB; non-trusted extensions (vector) pre-created as superuser.
Deploy-time: provision host + DNS/TLS.
Changelog — Admin
- feat(admin): deploy/admin base (Caddy+Postgres APP_DBS=twenty) + group-aware smoke/build
- feat(admin/twenty): Twenty CRM stack (server+worker+redis) on shared Postgres
- feat(admin): PostgREST + Adminer; admin spec + RUNBOOK section
- roadmap: add Admin group (Twenty CRM, shared Postgres) to backlog
Platform / shared (PR #2)
- refactor(postgres): group-driven APP_DBS + pgvector base
- feat(cli): bin/devtools single entrypoint (init / up|down|smoke|reporting --)
- fix(cli): trim devtools help to the comment block
- docs(recipes)+skill: capture recurring gotchas for future installs
- docs: essential repo + .claude markdown
- chore(conductor): add .conductor/settings.toml (setup/run/archive, nonconcurrent)
- decision: ingress = Caddy, not Traefik (record rationale + reconsider triggers)