Anomity

Anomity is an endpoint security and governance platform for agentic AI. A lightweight daemon on each managed device continuously inventories the AI tooling in use - MCP servers, coding agents, skills and instruction files, credentials, and account types - giving security teams visibility into a layer that sits between EDR and network controls. This connector lets you query that telemetry directly in Claude, without opening the console. What you can ask: Fleet overview - device counts by status (online, stale, offline) and open or critical finding counts across the fleet. MCP inventory - every MCP server detected across managed devices, grouped by trust level: official, community, or unknown-origin. AI accounts - which devices authenticate with corporate accounts versus personal ("shadow") accounts, plus data-sharing settings and BYOK configuration. Secret exposure - plaintext credentials found in AI configuration files, ranked by severity and pattern type. Capability surface - inferred permissions across MCP servers, flagging overprivileged devices and risky capability combinations. Skill and instruction-file risk - CLAUDE.md-class files and skills flagged by on-device content scanning for prompt injection, exfiltration patterns, and unsafe directives. Findings - policy violations filterable by status (open, acknowledged, resolved). Compliance status - readiness across SOC 2, ISO 42001, NIST AI RMF, and the EU AI Act, mapped to the controls each finding affects. Audit trail - 90 days of change history: every MCP server added or removed, every permission modified, on every device. Typical uses: investigating an alert without context-switching, preparing evidence for an audit, answering "which of our machines can reach production data through an agent?", or generating a weekly AI activity report. Anomity's approach is to govern rather than block - give teams the visibility to set informed policy instead of shutting down the tools developers are already using.