fix(agent-presets,connection): reclaim a deleted default, unpin a fence beside an open gate

Deleting the preset a user default names left the setting pointed at an id
nothing will ever supply again, and every session created without an explicit
pick then failed to start — the delete dialog called it 'new sessions cannot
select it', which understates a hard creation error. `remove` now clears the
user layer when it named the preset just deleted, exposing the deployment's own
default underneath. Storing a default that does not exist YET stays deliberate:
the roster is a live directory, so a name absent now may exist by the time a
session asks, and `resolve` still reports that case.

`agentPreset.select` also leaves the loopback set. It was pinned as a real
escalation — one preset mounts the toolset that edits the live runtime — but
`session.create` already takes an `agentPreset`, so pinning only the switch left
the same capability one method over. The deeper reason is that the capability is
not the preset's to grant: the deployment's own default already carries `bash`
and the filesystem tools, so any caller that may start a session at all can
already run commands as this process. `read`/`write`/`remove` stay pinned on
their own footing — those touch files, not sessions.

packages/host/apiproxy/README.md

@@ -36,7 +36,7 @@ Directory picking delegates to the composed `ctx.directoryPicker` backend ([the
 
 The `agentPreset.list` domain exposes the deployment's preset roster so a browser can offer a choice when starting a session; each row carries its `trust` (a `user` preset is exactly as privileged as the plugins it names) and whether it is the current default. A deployment composing no presets answers with an empty roster rather than an error, because sharing the host composition is a valid deployment. `agentPreset.select` recomposes one session's agent from a different preset, and is allowed only while the session is blank: once a turn has run, that history was produced under the preset's tools and swapping them would strand logged tool calls, so the attempt answers `agent-preset-locked`. The agent and the session survive — only the composition is swapped, and a failed swap restores the previous one.
 
-`agentPreset.read`, `write`, and `remove` author the compositions themselves. `read` reports the text with its `trust` and whether it is `writable`; `write` and `remove` refuse a preset that ships with the deployment, and `write` refuses an id that is not a containable directory name or text that is not a Cordis entry list (`agent-preset-invalid`), a refusal that reaches the caller as `agent-preset-read-only` for the shipped case. Together with `select`, these four are loopback-pinned in [`dsh-client-connection`](../../client/connection/README.md): a composition names the plugins a session runs, so reading one is reconnaissance and writing one is arbitrary capability. `list` stays ordinary — it carries ids and trust, and every preset picker needs it — and reports `authorable`, whether the deployment configures a root a new preset could be written to at all.
+`agentPreset.read`, `write`, and `remove` author the compositions themselves. `read` reports the text with its `trust` and whether it is `writable`; `write` and `remove` refuse a preset that ships with the deployment, and `write` refuses an id that is not a containable directory name or text that is not a Cordis entry list (`agent-preset-invalid`), a refusal that reaches the caller as `agent-preset-read-only` for the shipped case. These three are loopback-pinned in [`dsh-client-connection`](../../client/connection/README.md): a composition names the plugins a session runs, so reading one is reconnaissance and writing one is arbitrary capability. `list` and `select` stay ordinary — the roster carries ids and trust and every preset picker needs it, and choosing a preset grants nothing `session.create`'s own `agentPreset` did not, over a default that already carries bash. `list` reports `authorable`, whether the deployment configures a root a new preset could be written to at all.
 
 The `command.*` and `skill.*` domains expose the host command registry and skill catalog to clients. Every method addresses one session's agent by `sessionId` (a served session always has an Agent; `command.*` resumes cold sessions through the same path as `session.*`, while `skill.list` resolves the project root from the session header without touching the Agent registry). `skill.list` serves the browser's user-selected model-reference path, so it returns only skills that are both model-invocable and user-invocable; this domain has no direct skill-loading RPC. `command.execute` runs a slash-command line host-side with pure admission semantics: the response reports whether the line resolved to a handler plus the minted lifecycle `commandId` when it did (correlating the acknowledgment with the flow node), while the outcome rides the durably logged `command/run`/`command/done` lifecycle pair broadcast on the mux stream. Command handlers may legitimately outlast the 30-second transport health deadline, so `command.execute` carries only caller/connection cancellation; that signal cancels the running handler. `host/commands-changed` is the catalog invalidation frame: clients refetch `command.list` instead of diffing.