Group rows are not something to focus; the cards under them are one row
away. A collapsed group still stops the walk — its header is the only row
it has, so skipping it would leave the group unreachable.
Line 2 of a session card spelled out its blocked work ("1 uncommitted
file") and its command ("claude"). Both now render as marks: the
breakdown as diff notation, and the command's leading program as its
tool's icon in the web client.
The probe carries the line delta of the tracked changes — staged and
unstaged together, against HEAD — so `git add` never makes the counter
drop. Untracked files stay a file count; counting their lines means
reading every one off disk. A tracked change with no line delta falls
back to `~N` rather than claiming `+0 -0`.
The TUI keeps its command as text: a terminal cannot be assumed to have
a font carrying tool glyphs.
Removes the retained-worktree store end to end: the daemon's state-dir
store, GET/DELETE /api/worktrees, `maestro worktrees list|reclaim`, the
TUI overlay behind `w`, and the web `#/worktrees` view and header link.
The kill guard stays. A forced kill and a dirty natural exit now leave
the worktree on disk untouched instead of parking it in a store, so
removing one is an ordinary `git worktree remove`. A pristine tree is
still removed on exit. DELETE /api/sessions/:id always answers 204.
The header's debug button also goes; Ctrl+Shift+D and `?hud` still
toggle the HUD.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The worktree step ("+ new" plus the repository's existing worktrees) now
lives inside the collapsed "Advanced options" section in both the web form
and the TUI draft pane, so a plain spawn is command, project, name.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`r` opens a prompt pre-filled with the session's own name — empty when it
has none, so Enter cannot pin the command as the name. Enter saves, Escape
cancels, and an emptied field clears the name.
Refs #125
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Solarized (dark/light), Catppuccin Mocha, GitHub (dark/light),
Tokyo Night, Gruvbox, Nord, Dracula. Colours auto-tuned from the
canonical seeds to clear the palette contrast floors; tests now
assert every RGB preset, not just dark/light.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Every colour the TUI painted on its own chrome was a named ratatui
colour spelled out at the use site, so nothing could be themed and the
selected card had broken contrast: the highlighted card's black-on-cyan
base was patched with an activity colour chosen for the dark card
background, leaving AwaitingInput yellow on light blue.
Roles now live in `palette.rs`, named for meaning. A palette holds one
role set per surface — card and selection — plus the panel border in
its focused and unfocused forms, a focus accent, a group-row style for
the sidebar grouping, the status-bar mode indicators and error text.
Each surface carries its own activity colours, so the selected card's
dot keeps carrying state without wearing a colour picked for another
background.
Values are `Color::Rgb`, so contrast is asserted rather than left to
the terminal theme: every role clears WCAG AA on its own surface, and a
sidebar test fails if a card-surface state colour is painted onto the
selection background.
PTY cell colours stay a passthrough of what the program in the pane
emits — they are not chrome and not a palette role.
Line-1 actions anchor to the card's top-right corner on the highlighted
row only, reserving their width so the name truncates with an ellipsis
before it can ever reach them.
Reversed video made a highlighted card unreadable. Cards now sit on a
black card background painted edge to edge; the highlighted card swaps it
for black-on-cyan, matching the status bar's focus treatment.
Per docs/SIDEBAR.md: line 2 carries the activity dot + floored duration in
the state's colour, plus the blocked-work and exited fragments; it appears
only when time-in-state exceeds 30s, work is blocked, the session is dead,
or an Error state is in force. Blocked work moves off line 1. Card lines
truncate with an ellipsis instead of wrapping.
DirPicker mirrors the web filter/picker split: typing narrows and the
next Enter submits with the highlighted project directly (nothing was
chosen yet); an explicit Up/Down move onto a different row is
mid-selection, so Enter there only chooses it and a further Enter (now
the active pick) submits. WorktreeChoice, BranchExisting, Preset and
BranchMode already apply their highlight eagerly on Up/Down/Left/Right,
so Enter always submits there. Advanced and WorktreeToggle keep
activating on Enter, matching the web form's own button controls.
The draft pane's new-branch gutter was labelled "br:" in both branch
modes — since #94's web fix now says "New branch name" / "Pick a
branch", mirror the same distinction here: "new:" for the branch-name
input, "pick:" for the existing-branch list (which previously had no
label at all).
Keeps both #93 and #96 in the draft layout: the worktree toggle and body
stay behind the Advanced disclosure, while the worktree choice step
belongs to the directory picker (#95 rule 3) and stays visible once a
project is picked.
Closes#96
Mirrors the web project-picker redesign (#95) in the TUI (#96): step one
lists one box per repository (name, full path); picking one reveals step
two — "+ new" then that repository's worktrees newest-first, defaulting
to "+ new" except on a box that is itself a linked worktree. "+ new"
creates (cwd is the repo, create-worktree on); an existing worktree runs
in it (cwd is that worktree, create-worktree off). Neither step probes
/api/gitroot — only the free-text "Other directory…" hatch does.
Replaces the flat indented project+worktree list from #91/#64 with
projects::group_projects-backed DraftGroup boxes; the filter still
matches a repository via its own name/path or any of its worktrees.
Mirrors the web spawn form: worktree/branch controls sit behind a new
"Advanced options" disclosure line (bracket [+]/[-], toggled by Enter
or Space, closed by default) instead of always showing the checkbox
and branch rows. The header states whether a worktree will be created
so #91's off-by-default for a worktree row is visible without
expanding. Collapsed fields drop out of Tab order; expanding only adds
rows below the disclosure, never moving the ones above it.
Typing in the raw path field fired one /api/gitroot probe per
keystroke (#74). The draft now schedules the probe 250ms out — matching
GITROOT_PROBE_DEBOUNCE_MS in web/src/spawnForm.ts — and every keystroke
resets it; the 30Hz heartbeat polls for due probes so the burst's one
request fires after typing stops. The seq guard in resolve_gitroot is
untouched, so a late answer still cannot overwrite a newer input.
Picking a project row that is itself a linked git worktree used to
default 'Create git worktree' on, so spawning asked for a second
worktree of an already-checked-out branch and git refused. Both clients
now default the toggle off there: the session just runs in that existing
worktree with no worktree record (not adoption). Turning it back on by
hand still tries, per the issue's non-goal.
crossterm attaches KeyModifiers::SHIFT to every uppercase char, so a real
Shift+S delivered Char { c: 'S', shift: true }. canonical() turned that
into "shift-S", which the default "S" binding never matched, so
FastForwardSpawn was unreachable from the keyboard.
canonical() now drops the shift- prefix when the char is already
uppercase, since the case already says "shifted". Regression test drives
an actual crossterm KeyEvent through canonicalise() and the keymap,
rather than asserting on the binding string, per #70.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Session.blocked_work is Option<InTheWay>, refreshed by the daemon's
watcher; None for worktree-less sessions and until first probed.
Defaults on the wire so older payloads stay forward-compatible.
Nothing has honoured WorktreeSpec.cleanup_on_kill since #56, and after
#79 and #81 survival is decided by what is in a worktree, not by what
the spawn asked for. A field that lies is worse than a missing one.
Removes the field from the protocol, the --cleanup-on-kill flag from
the CLI, and the tests asserting its default.
The flag overrides the kill refusal (#81): the session ends, the guarded
worktree is retained, and the CLI prints where the work went — path,
branch, force-killed — instead of leaving the operator to wonder.
WORKTREES.md: write down that a forced kill of a tree whose guard is
clear removes it as usual — force changes what happens to blocked work,
not ordinary cleanup.
w opens the view; r confirms reclaim and says the branch survives.
The in-the-way summary uses InTheWay's Display impl so the TUI cannot
drift from the CLI's wording.
The cleanup guard plus deterministic paths had created a trap: one
untracked file an agent left behind — a log, a .env, a stray artifact —
made the guard refuse removal, the directory stayed, and every later
spawn of that branch hit PathExists. The branch was burned permanently
and the product offered no way to clear it. A leftover tree registered
for the same branch is now adopted, so the operator lands back in their
own unsaved work. A tree a *live* session holds is still rejected, since
two sessions must not share a checkout.
Also:
- Fast-forward still awaited /api/info, /api/projects and /api/names
inline, the exact freeze the draft loaders were restructured to avoid.
It posts to the same channel now.
- Draft updates were routed by tab index, which shifts when a tab closes:
an in-flight load could be dropped or land on the wrong pane. Drafts
carry an id and are found by it.
- extend_names left the name it had just displayed at the head of the
queue, so the next reroll produced it again and the key looked dead.
- The fast-forward fallback discarded info.home_dir, so its draft showed
unabbreviated paths and seeded "Other directory…" with / — disagreeing
with the normal draft entry point.
- Web fast-forward mounted via openSession, which replaces the focused
pane: pressing it while attached evicted that session from the layout.
It reuses an empty pane or splits, as the TUI does.
- branchEdited was inferred from the seeded branch value, but in "new"
mode that value is the auto-synced session name — so any form rebuild
froze the branch against later rerolls.
- resolveBaseRepo ignored the worktree checkbox, rebuilding the fieldset
expanded under an unticked box and probing gitroot for a worktree the
operator had turned off.
- short_hash used DefaultHasher, whose algorithm is not stable across
Rust releases, in a directory name — a toolchain bump would orphan
every worktree under the override root. Now FNV-1a.
The existing-branch picker could never succeed in either client: both
treated the main checkout's HEAD as available, and since that is usually
`main` it became the default pick — git then refuses the second checkout.
A TUI test had baked the bug in by asserting the selection lands on the
`is_head` entry. Both pickers now treat "checked out anywhere" as held.
Worktree paths were still not injective. Flattening `/` and `.` to `-`
made `feat/x`, `feat.x` and `feat-x` share one directory, and the
override root keyed on the repo basename alone, so `~/a/maestro` and
`~/b/maestro` collided. A branch name is already a valid relative path —
git stores it as refs/heads/feat/example — so it is used verbatim and
nests under `worktrees/feat/`; the override root gains a digest of the
full repo path. Git forbids holding both `feat` and `feat/x`, so the
nesting cannot collide for valid refs.
Also:
- Recovered sessions dropped their worktree association, so after a
restart a kill cleaned up nothing and the leftover directory blocked
that branch forever. It is persisted on Session now.
- The nesting guard compared git's resolved paths against an unresolved
one, so it silently passed on any symlinked repo path — and that guard
is what stands between a kill and an inner tree's uncommitted work.
- Draft loads ran on the TUI event loop: /api/projects on a cold cache
walks all of $HOME, freezing keys, frames and attached panes for up to
the request timeout. They are detached tasks posting to a channel now.
- Fast-forward in the web client created the session and never mounted
it, so `N` looked like a dead key while the TUI attached.
- The fast-forward fallback seed forced the worktree toggle off, against
both a normal draft and the TUI.
- The projects background refresh honoured the TTL, so two of every three
ticks were no-ops and requests paid for the rescan instead.
Miguel Palhas