`SuperHQ`: Run Coding Agents in `microVM` Sandboxes

Each agent gets a full Debian microVM, mounts your project, and writes through a tmpfs overlay so the host stays untouched. The diff-accept flow and out-of-sandbox API key handling make local agent runs much safer; worth testing now.

[ KEY POINTS ]

Every agent runs in its own full Debian VM, not a shared host process, so filesystem and process isolation are much stronger.
Project folders mount into the sandbox, but writes land on a `tmpfs` overlay first, letting you inspect changes before they touch the host.
Secrets stay outside the VM: API keys never enter the sandbox, which cuts one of the bigger risks in autonomous coding setups.
remote.superhq.ai adds remote control for workspaces and agents, so the local sandbox model can extend to anywhere access without changing workflow.

Originalgithub.com/superhq-ai/superhqRead original →

// related

#0001
#0001Agents & tools Google AI Forum2 hours ago
`Antigravity` `v2.0.0` Breakage: Staging Endpoint Misconfigured
50radar
AntigravityAI coding IDE — agent-based development workflows
Chat, completions, and agents can fail because the release points at a staging Cloud Code endpoint. Manual v2.0.1 reinstall is the practical fix when auto-update stalls.
- language_server.exe launches with --cloud_code_endpoint https://daily-cloudcode-pa.googleapis.com; production OAuth tokens get rejected there.
- Failures repeat every 25-30 seconds, breaking loadCodeAssist, fetchUserInfo, and fetchAvailableModels downstream.
- v2.0.1 reportedly clears the daily- endpoint, but some auto-updaters download to %LOCALAPPDATA%\antigravity-updater\pending\ without applying.
- Windows users stuck on v2.0.0 should skip debugging prompts and reinstall manually; this is a release config bug, not a project setup issue.
Source: discuss.ai.google.dev/t/fix-for-antigravity-v2-0-0-issueRead original →
FIG-0011:1
50radar
FIG-0011:1
#0002
#0002Agents & tools Google AI Forum2 hours ago
`Antigravity` users push back on hidden compute quotas
60radar
AntigravityAI coding IDE — agent workflow powered by Google models
Paid usage turned unpredictable after quota accounting moved from requests to hidden compute. A cheaper plan can now fail mid-workflow; budget risk matters more than model choice.
- A $20 Pro user reports 2-3 weeks of HTTP 429 lockouts despite visible quota remaining; reliability became the real blocker.
- The May 19 change replaced request limits with hidden compute-used accounting, making agent background scans and micro-queries harder to budget.
- Gemini 3.5 Flash is described as more verbose and weaker at coding, burning quota through long explanations instead of useful edits.
- Upgrade pressure jumps to $100-$200 Ultra or a reported 5-day ban after quota exhaustion; keep fallback IDE and agent paths ready.
Source: discuss.ai.google.dev/t/how-antigravity-became-cursor-2-Read original →
FIG-0021:1
60radar
FIG-0021:1
#0003
#0003Agents & tools Google AI Forum3 hours ago
`Antigravity` IDE and `AG2` Windows install paths collide
40radar
AntigravityAI coding IDE — Google's agent-style dev tool
The Windows installer can put both apps under the same Antigravity directory. Backup settings first; install order decides whether the IDE survives.
- Old IDE path and AG2 both target AppData\Local\Programs\Antigravity, causing the agent-only app to overwrite or mask the IDE install.
- Workaround: backup AppData\Roaming\Antigravity, uninstall AG2, install the IDE first so it creates Antigravity IDE, then restore settings.
- After the IDE launches once, rename the new roaming folder and move the old settings into Antigravity IDE; install order is the fix.
Source: discuss.ai.google.dev/t/fix-how-to-get-the-ide-back-and-Read original →
FIG-0031:1
40radar
FIG-0031:1

`SuperHQ`: Run Coding Agents in `microVM` Sandboxes

// related

`Antigravity` `v2.0.0` Breakage: Staging Endpoint Misconfigured

`Antigravity` users push back on hidden compute quotas

`Antigravity` IDE and `AG2` Windows install paths collide