plugin_grants_need_the_start_up_flag now drives `vectorcraft-cli mcp
--headless` over stdio. plugin.install with grant is refused without
--allow-plugin-grants and accepted with it.
A grant accepted under --allow-plugin-grants is written to grants.json like a
person's approval and counts as approved in later sessions and in the desktop
app. The CLI usage text, the plugin.install params text and docs/plugins.md and
docs/mcp.md said only a person's approval in the desktop app records one.
The fuzz test from #6223 passes grant to plugin.install, which a plain
session now refuses, so its cases stopped installing the module. The
docs line for a filter that lists inputs now says installing it needs a
person's approval.
Only a painted effect's references are checked, so a refusal recorded for a painted effect stayed in plugin.info after an effect that paints nothing replaced it under the same id.
plugin.install refuses grant from a request unless the session was started
to allow it, so the test from main that installs with grant turns that on.
The install dialog's comment names OK's Sense::click.
A painted live effect's pieces name swatches, so a kept evaluation could outlive a swatch that was renamed or removed and answer a click with art the document no longer draws. The kept evaluation now holds the document's swatches and color groups and is stale when they change.
A painted effect's cached result passed its swatch and pattern check and cleared the error of a failed run, so plugin.info reported none while the effect still failed on another object. A run and a reference check now record their errors separately, and each clears only its own.
tool.setOption on another tool reloaded every saved option of the active tool. During a shape drag that reload rewrote the options Escape restores, so Escape kept a side count an arrow key had changed. The active tool now reloads only the stores it shares with the tool that was set.
Kinsoku push-in room does not count a gap after inline art or inside a list marker, but the squeeze also closed those gaps. The squeeze now applies only to the gaps the room counts.
registry::remove released the plug-in set before it moved the revision, as install does not. A removal that finished just before another thread took the set could still move the revision while that thread held it, which made an_install_changes_the_revision_only_while_it_holds_the_set fail intermittently.