Antigravity

C▼ -14 vs last quarter

CAUTION: active unpatched backdoor + reliability crisis.

Rate this tool Add to compare

Dimension breakdown

Score · confidence

UX / DX

50% conf70

Capability

50% conf70

ReliabilityIncomplete data at this time

Value

50% conf60

CommunityIncomplete data at this time

Enterprise / Compliance

50% conf25

Autonomy

50% conf40

Integration

50% conf65

Sources blend review platforms, community sentiment, the Signal Radar and practitioner ratings. Scores re-blend each quarter.

Caution

active unpatched backdoor + reliability crisis.

Google's flagship agent-first dev platform — Antigravity 2.0 (GA at I/O 2026, May 19) is now a full agent control tower: standalone Desktop app, Go-based Antigravity CLI, Python SDK (self-hostable agents), Managed Agents via Gemini API, parallel subagents, scheduled tasks, 1M-token context, all on Gemini 3.5 Flash. Antigravity absorbs Gemini CLI (deprecated June 18) and Code Assist into one harness, sharply raising its strategic standing.

Security

Pillar Security RCE patched Feb 28 and FireTail's specific exfil path patched, BUT Mindgard 'Forced Descent' persistent backdoor remains UNPATCHED per primary source and the broad invisible-character exfil vector is 'won't fix' — critical-security-vuln cap maintained (Compliance=5).

Acute reliability crisis (chaotic 2.0 forced-update wiped IDEs/configs, agents crash mid-task, throttling, 'lobotomized' models) triggers a reliability-complaints cap (Autonomy=12). No Antigravity-specific SOC 2/SSO, no named enterprise customer, opaque credit pricing.

Not for production or enterprise use. Opus 4.8 model re-baseline.

Recommended

Use cases

Not yet assessed — this section fills in as ACES research covers the tool.

Score caps

Risk flags

Critical security vuln
securityTemporary
Unpatched critical security vulnerability
Caps Enterprise / Compliance at 25
Removed when — CVE patched and verified, or security incident resolved with post-mortem published
Reliability complaints
trustTemporary
Widespread reliability complaints (breaks often, unreliable output)
Caps Autonomy at 60
Removed when — 90+ days of improved reliability with community acknowledgment

Assessment

Status rationale

Detected status maintained. Cannot advance to Tracked because: (1) Mindgard Forced Descent persistent backdoor primary-source status still shows unpatched/unaddressed (global-config / trusted-workspace rules; survives uninstall/reinstall; no safeguard setting); (2) invisible-character source-code/credential exfiltration vector classified 'Intended Behavior (Won't Fix)' by Google (specific find_by_name path patched Feb 28); (3) no Antigravity-specific SOC 2 and no documented SSO/SAML/audit-log enterprise tier (Organization tier via Gemini Enterprise Agent Platform exists but is not yet a validated full enterprise offering); (4) acute reliability crisis around the chaotic 2.0 forced auto-update (editor/config wipe, agents crashing mid-task, throttling, 'lobotomized' models); (5) no named enterprise production customer; (6) opaque credit pricing (undisclosed credit-to-token conversion, ~92% free-tier quota cut).

Active capability development is strong and strategically central (Antigravity 2.0 platform, Gemini CLI consolidation into Antigravity CLI, SDK/Managed Agents, Gemini 3.5 Flash) — but security, reliability, and enterprise-tier blockers prevent enterprise recommendation. critical-security-vuln cap (Compliance<=5) and reliability-complaints cap (Autonomy<=12) both active; each caps the maximum signal level at Tracked.

Watch for

Movement triggers

Upgrade to Tracked: (a) Mindgard Forced Descent confirmed patched by Mindgard's own primary source AND the invisible-character exfiltration vector remediated rather than 'won't fix'; (b) Organization tier (Gemini Enterprise Agent Platform) validated as a full enterprise tier with documented SSO/SAML/audit logs; (c) 90+ days of reliability stability (Agent Terminated / quota-throttling crisis resolved and the post-2.0 rollout stabilized). Upgrade to Assessed: all Tracked conditions PLUS Antigravity-specific SOC 2 Type I certification AND hands-on internal evaluation completed.

Downgrade signals: confirmed in-the-wild exploitation of Forced Descent, additional data-loss incidents, Organization tier abandoned, Google product-rationalization signals targeting Antigravity, sustained Q3 reliability degradation, or the forum deprecation speculation becoming an official sunset.

Caution