Appsmith vs Tigris

Appsmith's recent release stream is dominated by security work. v1.99 alone landed roughly fifteen security-tagged fixes — multiple named GHSAs (super-user race condition, SSRF via send-test-email, OAuth2 callback ACL bypass, application snapshot delete permission, expanded metadata denylist), critical CVE patches (CVE-2025-70952, CVE-2026-33937 in handlebars, CVE-2026-22732 around Spring Security headers), AQL injection prevention in the ArangoDB plugin, and several reflected XSS and email-normalization fixes. The same pattern repeats in v1.98 (SQL injection in UQI filters, simple-git critical CVE), v1.96 (arbitrary file write outside repo scope, OS command injection in in-memory Git, XSS in Table HTML cells), and earlier. Feature work continues alongside but at a much smaller volume — Redis TLS, BetterBugs SDK, Favorite Applications V2, Helm extraVolumes.

The arc is clear: Appsmith is absorbing the output of what looks like a sustained external audit (or several converging ones) and using minor releases as the patch vehicle. The diversity of vuln classes across the ArangoDB plugin, Spring Security headers, OAuth2 callback, in-memory Git, snapshot deletion permissions, and metadata denylist points to a broad-surface review rather than a single component. Feature work isn't stalled, but it's clearly running second to the security queue.

Expect at least one or two more 1.9x releases to keep landing security patches before a 2.0 line emerges. Watch for a release that bundles fewer security items than features — that's the signal the audit cycle has caught up. Likely product-side bets are continued data-source TLS coverage and more granular permission scoping (the GHSAs around snapshots and OAuth2 lookup suggest the permission model is being tightened systematically).