v0.12.36: @newsroom.writer: A policy for every skill

A new article written here, from a voice memo of a conversation with a friend: his natural-language control policy for agent skills, a CI pipeline that checks every skill contributed to an organisation-wide repository against it, reading each skill in its context (trusted internal inputs against untrusted web content). Read as an agent behaviour policy for a skill (mandate, reach, gap); hope-based security and why it is still worth writing; an agent generating deterministic rules from the control while the description stays canonical; a catalogue of skills as one graph. Set against MongoDB's skill review guide, AWS's natural-language policy authoring and Simon Willison's lethal trifecta. His tool is described as he told it, not seen. Two figures.