Facio Blog

Practical notes on human-reviewed AI agents.

Payload-powered product notes, security writing, HITL patterns, and operational guidance from the Facio runtime: long sessions, Placet approvals, audit trails, memory, providers, channels, tools, and Docker-first operations.

Showing 66-70 of 235 articles.

Engineering

MCP Spotlight: Time MCP Server — The Anthropic Reference for Reliable IANA Timezone Math, Stopping the "Today's Date Is Tuesday" Hallucination Class

Jul 6, 2026Engineering

MCP Spotlight: Time MCP Server — The Anthropic Reference for Reliable IANA Timezone Math, Stopping the "Today's Date Is Tuesday" Hallucination Class

The official Time MCP Server by Anthropic — two tools (get_current_time, convert_time), IANA timezone database, DST-aware, ISO 8601 output, MIT-licensed. The ground-truth time primitive that eliminates the "today is Tuesday" hallucination class. Available via uvx, Docker (mcp/time), or pip.

Product

Facio's Policy Engine: How AI Agents Stay Aligned With Business Rules Even When the Model Wants to Stray

Jul 6, 2026Product

Facio's Policy Engine: How AI Agents Stay Aligned With Business Rules Even When the Model Wants to Stray

An AI agent that follows business rules is valuable. An AI agent that improvises around business rules is a liability. The naive approach to alignment — putting rules in the system prompt — fails because the model can be social-engineered, distracted, or simply wrong about when a rule applies. Facio's policy engine is a runtime enforcement layer that makes business rules architecturally binding: every agent decision is checked against the policies before execution. Rules can't be bypassed because the runtime refuses to execute non-compliant actions.

Human-in-the-loop

HITL Observability: Why the Agent Must Be Transparent to the Reviewer (and Not Just to the Developer)

Jul 6, 2026Human-in-the-loop

HITL Observability: Why the Agent Must Be Transparent to the Reviewer (and Not Just to the Developer)

Most agents are transparent to developers — trace logs, reasoning chains, debug panels. But they're opaque to reviewers, who see only the final action proposal. Reviewers make worse decisions on opaque agents — they rubber-stamp what they can't evaluate. The HITL system needs its own observability layer, designed for the reviewer's eyes, not the developer's.

Engineering

MCP Spotlight: Fetch MCP Server — The Anthropic Reference for Safe HTML-to-Markdown Web Fetching With URL Allow-Lists and the Read-Only Default

Jul 5, 2026Engineering

MCP Spotlight: Fetch MCP Server — The Anthropic Reference for Safe HTML-to-Markdown Web Fetching With URL Allow-Lists and the Read-Only Default

The official Fetch MCP Server by Anthropic — one tool (fetch), read-only by design, HTML→Markdown conversion with raw/JSON modes. MIT-licensed, available as NPX package or Docker image. The read-only web access default for AI agents in 2026.