Keep the personal half of the desktop in one place, and ask before installing it

Agent instructions, skills, SSH host aliases and expansion triggers are worth
having identical on every machine one person owns, and belong in none of the
shared configuration. They live in user/ now, with a manifest saying where each
piece goes and a link-user stage that puts it there.

That stage does nothing unless the machine said yes. Somebody who clones Panama
to try the desktop keeps their own ~/.claude/CLAUDE.md exactly where it was;
the question names the destinations and defaults to no. Anything displaced goes
to config/old rather than being deleted.

~/.claude/CLAUDE.md and ~/.codex/AGENTS.md were byte-identical copies of one
file, which is the drift this exists to prevent.

Also adds the vitals toggles for the battery and Claude usage readouts, which
had preferences and no way to reach them.

user/agents/skills/research/SKILL.md

@@ -0,0 +1,36 @@
+---
+name: research
+description: Use when a question needs reading legwork against primary sources and the findings should be written down, such as how a third-party API behaves, how a protocol works, or how a vendor compares. Not for library API syntax, which is find-docs.
+---
+
+# Research
+
+Delegate reading to a background agent so the main thread keeps working, and land the findings as a
+cited Markdown file.
+
+**This is not the skill for library or framework API questions.** Version-specific syntax,
+configuration options, and SDK signatures belong to `find-docs`, which queries Context7 directly and
+is faster and more accurate for exactly that. Reach for `research` when the question is *not* a
+documented library API: how a third-party service actually behaves, how a protocol or file format
+works, how two vendors compare, what a spec requires.
+
+## Process
+
+Spin up a **background agent** so you keep working while it reads. Its brief:
+
+1. **Investigate against primary sources**: official documentation, the source code itself, the
+   specification, the first-party API. Not a secondary write-up of one. Follow every claim back to
+   the source that owns it, and where a claim only exists in a blog post, say so rather than
+   promoting it.
+2. **Write the findings to a single Markdown file**, citing the source for each claim as a link.
+3. **Save it where this repo already keeps such notes.** Match the existing convention; where there
+   is none, use `.claude/docs/research/<topic>.md` and say where it landed.
+
+Where the question has an answer that is genuinely contested or version-dependent, the file records
+the disagreement rather than picking a winner silently.
+
+## Done when
+
+- Every claim in the file carries a link to the primary source that owns it.
+- Anything that could only be found in a secondary source is labelled as such.
+- The file's location was reported.

user/agents/skills/research/agents/openai.yaml

@@ -0,0 +1,3 @@
+interface:
+  display_name: Research
+  short_description: Background agent investigates against primary sources and writes cited findings.