// observability + sovereignty for the agent era

The lab between
your machines and the cloud.

AI agents read your files, hit your network, and act on your behalf — most of the time invisibly. 127.0.0.1 labs builds the local-first sentinels that watch and gate them for you. Endpoint to edge. Watch. Classify. Gate. Just signal.

See rAIzr → See Mimir → Notes →

[01]The visibility gap

Every modern endpoint runs at least one autonomous agent — Claude, Copilot, ChatGPT, Cursor. They have access to your files, your keychain, your network. Almost nobody can see what they actually do.

88%

of organizations had an
AI-agent security incident in 2025

21%

have full visibility into
what their agents access

1.8k+

internet-reachable MCP servers
running with no authentication

66%

of scanned MCP servers had
security findings (1,808 sample)

// sources: Gravitee — State of AI Agent Security Report · Akto — 2025 State of Agentic AI Security · Red Hat — MCP Security: The Current Situation · AgentSeal — We Scanned 1,808 MCP Servers.

[02]The lab

What we're building

Two products today, more on the way. Each one solves a specific gap in how organizations observe what AI is actually doing on their infrastructure. Pick one to dive in.

● ALPHA

rAIzr

Raise the bar, cut the risk.

Runtime-aware AI Zero-trust Response. Per-process attribution via ESF. Local SQLite, encrypted at rest. Live menubar status. Zero cloud.

Read the spec →

● v0 · pivot in progress

Mimir

See the signal, know the pattern.

Multi-source Intelligence Monitoring & Incident Response. SecOps dashboard + OSINT aggregator. 100% local. Four-node topology, SQLCipher, audit-logged everything.

Read the topology →

○ THE LAB

More to come.

Other tools, in other lanes.

We build observability that we ourselves need. When the next gap appears — and it always does — we'll surface it here.