docs: Sf-Workspace, and what the gateway sets (#90) README, AGENTS.md, the skill and the wiki name the Sf- headers, say the gateway sets Sf-Workspace and that Claude.ai never sends it (ruling 12), and stop saying the session names are refused (ruling 10, revised). The CHANGELOG's [Unreleased] gains the BREAKING line, and its workspace line no longer contradicts it. From the whole-branch review: two test names and a heading say Sf-, and a test pins ?session= beside an old header. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent: done
test: capture joining a hold opened for control, and the stop said as it is The whole-branch review found the monitor's path where capture joins a connection held only for control untested; it now has tests, with the reverse and the lapsed-capture path. AGENTS.md, the skill and the await_control notes no longer promise every waiter a stopping answer, since uvicorn's graceful timeout usually cancels it first, and say that a read begun before Control was asked can be refused part-way. live/pages.py imports the BiDi connection for typing only. Agent: done
docs: control and await_control for agents and maintainers The skill's workspaces reference says what a refusal looks like, what the control field on a result means, which reads still answer, and how await_control waits and what it returns. AGENTS.md replaces the live view's "nothing is refused" with the rules for a person in Control, and brings the lock, capture, monitor and scaling sections up to date. The wiki gains await_control's notes, and the changelog its line.