Cut every comment in the CLI to two lines (GRYT-1056)

.github/workflows/bump-gitlink.yml

@@ -1,11 +1,7 @@
 name: Bump superproject gitlink
 
-# Tells Gryt-chat/gryt that main moved, so it can fast-forward this
-# repository's gitlink instead of waiting for its hourly sweep. Implementation
-# lives in Gryt-chat/.github so all repos share one copy.
-#
-# Needs the org secret GITLINK_DISPATCH_TOKEN. Without it the dispatch is
-# skipped with a warning and the sweep picks the change up instead, later.
+# Tells Gryt-chat/gryt that main moved so it can fast-forward the gitlink. Shared
+# implementation in Gryt-chat/.github; needs the org secret GITLINK_DISPATCH_TOKEN.
 
 on:
   push:

.github/workflows/ci.yml

@@ -20,3 +20,10 @@ jobs:
       - run: go test ./...
       - run: go vet ./...
       - run: go build ./cmd/gryt
+
+      # No comment runs past two lines. The script carries the list of paths still
+      # to be swept, and that list only ever shrinks.
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+      - run: node scripts/check-comment-length.mjs

.github/workflows/discord-ci-notify.yml

@@ -1,12 +1,7 @@
 name: Discord CI notification
 
-# Reports failed workflow runs to #ci in Discord, and the recovery when one
-# goes green again. Implementation lives in Gryt-chat/.github so all repos
-# share one copy. Needs the org secret DISCORD_CI_WEBHOOK_URL.
-#
-# Workflows are listed by name rather than left to a wildcard: GitHub's docs
-# don't say whether omitting `workflows` means all of them, and an unlisted
-# workflow notifies nobody. Add new workflows here as you add them.
+# Reports failed workflow runs to #ci in Discord, and the recovery. Shared implementation
+# in Gryt-chat/.github. List new workflows here — an unlisted one notifies nobody.
 
 on:
   workflow_run:

.github/workflows/release.yml

@@ -1,14 +1,7 @@
 name: Release CLI
 
-# Dispatched by hand, same shape as the SFU release. Version comes from the
-# tags rather than a committed file, so nothing is pushed to main and this
-# works against a protected branch.
-#
-# goreleaser does the building and creates the GitHub release itself, so there
-# is no separate action-gh-release step here: the archives, the checksums and
-# the release notes all come out of `goreleaser release`. install.sh in the
-# superproject reads those assets, so the archive naming in .goreleaser.yml is
-# load-bearing — changing it breaks the one-line install.
+# Dispatched by hand, same shape as the SFU release: the version comes from the tags, so
+# nothing is pushed to main. install.sh reads the archive names in .goreleaser.yml.
 
 on:
   workflow_dispatch:
@@ -148,12 +141,8 @@ jobs:
             esac
           fi
 
-          # Until now the channel input was validated, written to an output and
-          # read by nothing. goreleaser decides prerelease from the tag alone,
-          # so dispatching with beta still shipped a stable release that
-          # /releases/latest returned and the installer picked up. Both
-          # install.sh and `gryt update` resolve /releases/latest, which
-          # excludes prereleases, so the distinction has to come from here.
+          # goreleaser decides prerelease from the tag alone, and both install.sh and
+          # `gryt update` resolve /releases/latest, which excludes prereleases.
           if [[ "$CHANNEL" == "beta" ]]; then
             BASE="${NEW_VERSION%%-*}"
             NEXT=1

.github/workflows/vikunja-task-done.yml

@@ -1,9 +1,7 @@
 name: Sync Vikunja task
 
-# Keeps the Vikunja task named in the branch (e.g. claude/GRYT-1-slug) in step
-# with the PR: labelled "in review" and linked when it opens, marked done when
-# it merges. Implementation lives in Gryt-chat/.github so all repos share one
-# copy. Needs the org secret VIKUNJA_API_TOKEN.
+# Keeps the Vikunja task named in the branch in step with the PR: in review when it opens,
+# done when it merges. Shared in Gryt-chat/.github; needs VIKUNJA_API_TOKEN.
 
 on:
   pull_request:

cmd/gryt/main.go

@@ -58,10 +58,10 @@ func main() {
 	}
 }
 
-// runDoctor prints every check and returns the exit code, so that a script can
-// gate on it.
-// runUpdate replaces this binary with the newest release, or with --check only
-// reports whether there is one.
+// runDoctor prints every check and returns the exit code, so a script can gate on it.
+
+// runUpdate replaces this binary with the newest release, or with --check only reports
+// whether there is one.
 func runUpdate(args []string) int {
 	checkOnly := len(args) > 0 && (args[0] == "--check" || args[0] == "check")
 

internal/app/entries.go

@@ -2,13 +2,8 @@ package app
 
 import "github.com/Gryt-chat/cli/internal/config"
 
-// An entry is a row in the table. Most are servers; the last two are the
-// pieces the machine runs one of and every server shares.
-//
-// Before this the model assumed every row was a config.Profile, which is why
-// the SFU and the object store appeared nowhere: the dashboard could say
-// "voice server is not running" and then offer nothing to do about it, and
-// there was no way at all to read the SFU's log.
+// An entry is a row in the table. Most are servers; the last two are the pieces the machine
+// runs one of and every server shares, which used to appear nowhere.
 type entryKind int
 
 const (
@@ -28,12 +23,8 @@ type entry struct {
 	role string
 }
 
-// entries lists the servers, then the shared pieces.
-//
-// Image workers are deliberately absent. One runs beside each server inside
-// that server's own compose project, so `docker compose logs` for the server
-// already carries its output; giving it a row would add a line to look at and
-// no information that is not already one keypress away.
+// entries lists the servers, then the shared pieces. Image workers are absent: one runs
+// inside each server's own compose project, so its output is already in that server's logs.
 func (m Model) entries() []entry {
 	list := make([]entry, 0, len(m.profiles)+2)
 	for _, profile := range m.profiles {
@@ -57,10 +48,8 @@ func (e entry) key() string {
 	return e.container
 }
 
-// actions are the things worth offering for a row in its current state.
-//
-// Start on a running server used to be accepted, run `compose up` again, and
-// report "Start X" as though something had happened. The state decides now.
+// actions are the things worth offering for a row in its current state. Start on a running
+// server used to be accepted and report success without anything happening.
 type actions struct{ start, stop, restart bool }
 
 func availableActions(running, unknown bool) actions {