Skip to content

Finders

This page lists the Jump All modal, jump back, and the Node Finder.

Jump All Modal

Press ` (backtick) on any tab to open the Jump All Modal. It displays all entries across Agents, Artifacts, and Services tabs with the same adaptive one- or two-character hints used by current-tab entry jump. Completing an entry's hint switches to the appropriate tab and focuses it.

Up to 62 entries use 0–9, a–z, A–Z. Larger result sets use fixed-width pairs from 00 through ZZ; a first character is consumed without closing the modal, and uppercase characters remain case-sensitive.

Key Action
Hint Jump to the corresponding entry
` Jump back to the previous position (see below)
Ctrl+D / Ctrl+U Scroll the entry list down / up
Esc Close modal (any other key that is not a hint also does)

The modal groups entries by tab (Agents, Artifacts, Services) and shows contextual information for each: PR names and statuses, agent names with running indicators, and Axe routine/command labels.

Jump Back

Both jump modals support a jump-back feature for toggling between two entries:

  • Backtick jump-back: Pressing ` inside the Jump All Modal returns to the previous position, enabling quick toggling between two entries across tabs.
  • Apostrophe jump-back: Pressing ' twice ('') in the single-tab entry jump mode jumps back to the previously jumped-from entry. The footer shows a "JUMP" mode indicator with ' back when a target exists.
  • Fast jump: Ctrl+O runs the same current-tab jump-back path without painting hints first; when no jump-back target exists, it selects the first current-tab hint.
  • Forward jump: After walking backward, Ctrl+Shift+O walks forward through that current tab's jump stack. Agents, Artifacts/Patches, and Services keep independent back and forward positions.

The single-tab variant (' apostrophe) shows entries only from the current tab with the same hint-character navigation.

Node Finder

Press " (quotation mark) on the Agents tab to open the Node Finder modal. It lists every reachable sase node as a tree — clans, agent nodes, session turns (including monitors and gates), workflow roots, workflow agent steps, and stand-alone named procs — including rows hidden by collapsed folds, collapsed grouping banners, collapsed or isolated tribe panels, and the Agents query. Jumpable live node rows carry jump hints; Archive results below them use selection and Enter.

Key HINTS mode (default) SEARCH mode (query focused)
0-9a-zA-Z Complete a hint and jump, or set a pending prefix Edit the query; live refilter
Tab / Shift+Tab Focus the query Back to HINTS, query kept
/ Focus the query Types /
Enter Jump to the highlighted node Jump to the highlighted node
Ctrl+N / Ctrl+P, ↓/↑ Next / previous jumpable row (wraps) Same
" Jump back (same as Ctrl+O) Types "
Backspace Cancel a pending prefix; otherwise swallow Delete a character
Esc Cancel a pending prefix; otherwise close Back to HINTS, query kept
PgUp / PgDn Scroll the preview pane Scroll the preview pane
Any other key Swallow; the footer flashes no hint ‹x› Types

The why-hidden glyph legend: blank means visible, ◆ you are here, ⊘ hidden by the Agents query, ◌ hidden by I (until listed, shown as an I hides K chip), ▭ inside a collapsed or isolated-away tribe panel, ≡ inside a collapsed grouping banner, ▸ inside a collapsed fold.

"Hidden" covers folds, banners, panels, and rows the Agents query hides. Jumping to a query-hidden node clears the Agents query (recorded in query history, announced with a toast); restore it with the query-history keys. The one-line jobs of the sibling surfaces: ' jumps by hint on the current tab, ` jumps across all tabs, digits jump roster members, and " in the finder jumps back.

A ◷ In Archive section follows the live rows. It lists up to five dismissed runs from this machine: the first five names in the Archive's day order when the query is empty, or the five best fuzzy name matches once you type. While that list is loading the section reads Archive warming… or Searching Archive…; an empty search result reads No archived name matches., and a loading or search error reads Archive unavailable. The finder searches an already-warm Archive snapshot; if none is available when it opens, the section fills in once the Archive warms while the finder stays open (it rechecks on each keystroke and for about 15 seconds after opening). Select a result with the arrow keys or Ctrl+N / Ctrl+P. Its preview reads ARCHIVED RUN, Dismissed on this machine · read-only, and ⏎ opens the run in the Archive. Enter switches to the Archive and selects the run without restoring it. This replaces the Archive query with in:archive; it does not add a name filter. The toast reads Node Finder · opening archived run · read-only. Archive results carry no jump hints.

The preview sits beside the list. On a terminal at least 140 columns wide the list also keeps a status column. From 100 columns up to that width the two panes split evenly and the status column is hidden. Below 100 columns the preview stacks under the list. PgUp and PgDn scroll the preview in both modes.

The preview paints from the snapshot taken when the finder opened. The highlighted row shows its kind and name, a compact identity block, and a breadcrumb from the tribe panel through ancestor names to the row (@default ▸ parent ▸ name). It also says whether the row is visible, and what Enter will do: ⏎ selects it when the row is already visible, or the reveals it will perform first (clearing the Agents query, opening a hidden panel, opening a collapsed group, expanding folds) and then selecting it. A session row lists its turns. A clan lists a status tally and up to twelve members, then … N more members when the clan is larger. A monitor or named proc shows its command and an output tail. A gate turn shows its label and state. A workflow root shows a step-status tally. A non-node row says Select a node to inspect it.

Agent turns then fill two more sections from that turn's artifacts: a PROMPT head and a REPLY · tail. A session container uses its newest agent turn for those sections and labels the prompt with that turn's name. Clans, monitors, gates, and named procs stop after the in-memory preview. The prompt and reply slots read ⋯ loading until that read finishes, then (no prompt recorded) or (no reply recorded) when the artifact is empty. The reply tail names how many earlier lines were left off.

_[project]: A sase project is a named unit of work registered with SASE. _[workspace]: A sase workspace is a numbered clone of a project's primary repo, managed by the workspace store and tracked in that project's registry.json. _[agent]: A sase agent is an agent session or a single agent that does not belong to a session. _[prompt]: A prompt is the text SASE hands to a provider to start an agent turn. _[macro]: A macro is a reusable prompt triggered with #name in agent prompts. _[Patch]: A Patch is SASE's local unit of change. _[bead]: A bead is one tracked work item in SASE's git-native issue tracker. _[plan]: A plan is a persisted, approved artifact recording the intent behind agent work. _[TUI]: The TUI is SASE's terminal interface for running agents and reviewing results. _[epic]: An epic is an executable multi-phase plan run with sase bead work. _[gate]: A sase gate is a durable, command-backed request for one user decision, created when work must pause for the user. _[monitor]: A sase monitor is a monitor turn that runs one long command under a detached supervisor, so the command outlives the agent that started it. *[stitch]: A stitch is the lightweight ordered change record inside a Patch's STITCHES: section.