Detectify launches MCP Server to secure AI coding loop. Detectify exposes its scanners via the MCP standard so AI agents can trigger validation scans, generate patches, and receive structured remediation tasks. This turns security tooling into first-class agent-callable services, letting you bake automated validation and repair into delivery lanes (Principles 02 & 09).

This Half-Gigabyte AI Model Runs Local Agents on Your Phone. OpenBMB’s MiniCPM5-1B runs MCP-enabled local agents on smartphones with 128K context, enabling offline tool use and long-context workflows despite some reasoning failures. Expect outcome engineering to split between cloud-backed and edge-first agents, forcing new tradeoffs in latency, privacy, and resilience (Principles 07 & 06).

OpenRouter Raises $113 Million for Model Marketplace. OpenRouter is building a multi-model inference marketplace with routing, billing, and governance primitives to orchestrate enterprise models at scale. Use model routing and marketplace governance to manage cost, compliance, and fallbacks across heterogeneous agent stacks (Principles 09 & 10).

Observability Is Your Profit Center Now — Honeycomb’s Christine Yen. Honeycomb argues production signals should be compiler inputs for autonomous agents, reframing observability from risk control to active input for agent decisioning. Instrument telemetry as first-class inputs to agent loops so agents act on grounded, auditable signals rather than brittle heuristics (Principles 02 & 06).

Anthropic Adds 28 Security Integrations for Claude Governance. Anthropic lets enterprises route conversation content and activity telemetry into DLP, SIEM, identity, and observability tools, creating richer audit trails and enforcement points. That integration reduces blind spots in deployed agent behavior and makes governance, logging, and compliance practical at scale (Principles 10 & 13).