fix(chat): align agent message expansions to the transcript grid

Merged
TE
4d ago
+8-3

The opened agent-message body and delivery details used a local 22.5px indent, so message rows could drift from the transcript grid when its glyph slot changes. This moves that indent into the shared transcript token and keeps the existing 22.5px geometry at today's defaults.

Leo said: "I need to, like, see the messages coming in back and forth at the very least."

Gallery stills (2x, synthetic sent-message fixture; the token preserves the current pixels):

Design reference: design/chat-transcript.md requires the shared transcript grid and forbids local row geometry.

Risk: low. This is a CSS token refactor; the default computed indent is unchanged. Independent visual review should inspect the after light and dark stills.

agent-identity footer --repo manaflow-ai/cmux


View with [code]smith Autofix with [code]smith
Need help on this PR? Tag @codesmith-bot with what you need.


Summary by cubic

Aligns agent-message expansions to the transcript grid by replacing the hardcoded 22.5px local indent with a shared token computed from the glyph slot and inline gap. The default computed indent stays at 22.5px, so there's no visual change at today's defaults, but message rows no longer drift when the glyph slot changes.

The token lives on the transcript grid, so the message card and delivery details follow the row instead of carrying a second local pixel indent. This is a CSS token refactor with unchanged default geometry; no migration or rollout steps needed.