Port of Context launches MCP context layer for regulated firms
Start here
Port of Context has launched a self-hostable MCP context layer aimed at regulated industries. It aggregates existing MCP servers behind a single endpoint and exposes their tools to agents as code, with each tool call running in an isolated environment by default rather than reaching directly into customer systems. The company says any existing MCP server works unchanged. The platform can run in a customer's own cloud account, in their data center or on an air-gapped network, and supports Claude, GPT, Gemini or open-weight models. Its core is open source under the MIT license, so it can be read, run and self-hosted without a commercial agreement, and every run is traced under its open AVP observability spec. Port of Context cites an Agentic AI Foundation case study in which each run fell from 93.6k to 31.8k input tokens at half the cost. That figure comes from the cited study, not from independent testing.
Generated from reports · updated 1 h ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AIPort of Context launches a token-efficient MCP context layer for regulated firms
Port of Context launched a self-hostable MCP context layer aimed at regulated industries, aggregating existing MCP servers behind one endpoint and exposing their tools to agents as code. Its code mode cites an Agentic AI Foundation case study in which each run fell from 93.6k to 31.8k input tokens at half the cost, and the platform runs on Claude, GPT, Gemini or open-weight models in a customer's cloud, on-prem or air-gapped network, with the core MIT-licensed and every run traced under its open AVP observability spec.
Heat for this event
Not enough continuous observations yet to draw a trend.