What the internet is saying about Supabase

What the internet is saying about "Supabase backend"

🟢 Supporters

• •Supabase is being picked as the default BaaS in agent and factory tooling. One issue titled "BaaS pack — Supabase first, Firebase as the alternative" makes RLS and security rules a gate from the first migration [7]. That treats Supabase's security model as a baseline to build on.
• •Supabase is moving into compute and scaling. Supabase announced "Supabase Compute" in private alpha [1]. Deepthi Sigireddi of Supabase was interviewed on Vitess and Multigres [2]. Both suggest the platform is expanding beyond a plain Postgres host, though the evidence gives little detail on either.
• •Many real projects build on it. Supabase schema + RLS, migrations and Edge Functions appear across unrelated repos, including a Whop/Cloudflare app [3], a Syra function PR [10], a rentals app that recovers its Edge Function source into GitHub [12], and a client portal [14]. The sheer spread of use is a signal in itself.
• •An ecosystem of tooling is forming. "Sodabase" launched on Show HN to find and fix broken Supabase tables [4].

🔴 Skeptics

• •AI-agent guidance for Supabase may not help. A study found Supabase's official Postgres skill "took longer than no skill" [6]:
  • •Claude Code / Sonnet: time +7.6%, tokens +17.7%.
  • •Codex: time +11.8%.
  • •Claude Code / Haiku: time +6.2%, tokens +14.1%.
  • •Only Haiku + Copilot improved with the skill [6].
• •Operational risk shows up in production-readiness work. A P0 prelaunch blocker cites a staging backup/restore and migration-branch parity failure, with a warning not to apply migrations to live Staging or production before backup/restore [5]. A related final go/no-go thread keeps protected staging SQL and production on HOLD [11].
• •Platform lock-in is a live concern. A repo called "unvendor" asks for migration stories from people who moved off Supabase, Firebase and similar BaaS platforms, or tried and gave up [9].
• •Broken tables are common enough to need a tool. The existence of Sodabase implies people hit this problem [4].

⚪ Neutral Observers

• •Teams compare options before committing. One issue asks a dev to compare Supabase, Firebase and Cloudflare D1/R2 on free-tier limits, cost and operations [8]. Another frames Supabase and Firebase as alternatives "under stated conditions" [7].
• •Operational hygiene is an ongoing task. One PR trims scheduled-task logs to seven days of history while archiving failures, to limit disk/WAL use [15]. Another fixes a 401 where a service key was sent to Kong's /rest/v1 [13].
• •Auth is a recurring need. An app is exploring user accounts and registration [18], and another decouples frontend auth [29]. The evidence does not say Supabase Auth is used in either.
• •Many mentions are incidental. Several listed items (e.g., [34], [35], [36]) give no substantive Supabase signal.

Cross-Platform Signals

• •GitHub dominates by volume. Most items are issues and PRs where Supabase is part of the stack, not the subject. Several have long threads, such as 52 comments on the staging backup/restore blocker [5], 55 on the Kong key fix [13], and 108 on a master TODO [17].
• •Hacker News is small but more opinionated. Items there have 3–4 points and 0–1 comments [1][2][4][6]. They cover Supabase's roadmap, a tooling gap, and a critical benchmark.
• •There is no strong consensus or open conflict. GitHub shows steady adoption, with caution around backup, migration and security gates. Hacker News shows interest in new capabilities alongside skepticism about AI-skill effectiveness.

Community Verdict

Supabase looks like an established, default-choice backend for small and mid-sized projects, especially those built with AI-assisted workflows. Developers treat RLS, migrations and backup/restore as serious work that needs explicit gates [5][7][11]. Supabase is also pushing into compute and scaling [1][2]. The critical points are narrow rather than hostile: a benchmark showing its Postgres agent skill added time [6], migration and recovery friction [5], and interest in escaping BaaS lock-in [9]. Engagement on the Hacker News items is low, so these conclusions rest on thin evidence.

---

Based on Hacker News + GitHub data from the last 30 days, researched with last30days by Matt Van Horn.