Desktop app

archeus --gui.

Everything the terminal UI does, as a native desktop app — full feature parity, served locally (loopback-only, works offline). No Python dependencies; the browser bundle vendors three.js and anime.js (both MIT, served from /vendor/, never a CDN).

• Shells — PyQt6 native window if installed, else an Edge app-mode window, else the browser (gui_shell setting: auto / qt / edge / browser). The bottom-left toggle (or ui_mode) picks which interface starts by default; --tui/--gui always override.
• Open project by path — type or complete any folder and it opens as a project page: its sessions, memory, CLAUDE.md and the rest, whether or not it has ever had a session. Launching is the page's own button, the same as for any other project.
• Projects & sessions — sidebar with live filter and quick-resume; per-session resume / fork / rename / tag / archive / restore / delete / export markdown / transcript with session info / changed files.
• Launch modal — effort, model, permission mode, account, thinking cap, subagent model, session name, worktree — as one-click chips, prefilled from your defaults. Sessions open in a real new console window.
• Sidebar — five sections, each a set of sub-tabs across the top of the page: Context (global CLAUDE.md, output styles, MCP servers), Library (agents, skills, hooks, plugins), Activity (usage, loops, logs), Harnesses (every CLI archeus drives, its accounts and Claude Code's own record of itself) and Settings (launch, appearance, paths & limits, models, updates). Search is the box on the dashboard and Ctrl+K; Help is the ? at the foot of the sidebar. The same five names, in the same order, are the terminal UI's five submenus and the five sections of this documentation.
• Project tabs — four tabs, each grouping the pages that answer one question: Session, Context (memory, CLAUDE.md, audit), Project (tools, usage, repos) and Actions (plan → execute, code review). Each group opens on the tab it is opened for, and remembers which one you were last on. The page you are on names itself in one line under the tabs, from the same sentence the help page renders. The three that carry the most:
  • Memory — what archeus knows, and what it costs. It opens on what it knows, then what it learned, then the price. An inventory of every memory artifact the project has, each with what writes it, whether it reaches a session always / lazily / per prompt / never, its size, and its action — plus the sentence the UI never used to say: ~N tok read on every turn, ~M tok across the rule files that cost nothing until Claude opens a matching path. Then the live state: what the last cycle extracted and what it spent, total spend to date, what is queued, what eviction dropped, which entities recall reinforces most, lessons with how many sessions each is from being dropped, and workspace health as per-check rows with the button that fixes each one. Build / ask / recall preview / lessons review, with live scan progress and the outcome of the last run.
  • CLAUDE.md — the file, block by block. It opens on Your prose: an editor holding only the half of the file you wrote, with a Save button. Everything archeus generates is left out of that box and put back untouched on save, so there is nothing there you can break, and KEEP fences travel with your text. Below it: your prose, KEEP-fenced regions, AUTOGEN, SESSIONS and the MEMORY digest, each with its token cost and the one button that regenerates just that block; the memory files map with broken @imports flagged; and every version archeus replaced, with diff and restore.
  • Audit — what one turn costs across every surface at once, project-scoped and account-scoped together (global CLAUDE.md per account, system-prompt file, SessionStart hooks, MCP schemas), plus deny rules for token-heavy paths, project health with the button that fixes each issue, and Claude Code's own record of the project.
  • Also: Tools in two columns — what the project launches with (its agents picker, --add-dir directories, extra PATH entries) on the left, what it has been doing (most-run commands, what to work on) on the right — plus Usage, Repos, Plan → Execute (plan model + effort, execute via Anthropic or free OmniRoute, full explanation inline) and the architecture Graph, which opens in its own window from the tab strip.
• Session rows — every session of the project across every account. At rest the row is its title: the whole width, with the full text as a tooltip when it is long. Hover (or tab into it) and the actions fade in over the right end without moving anything — transcript, export, changed files, checkpoints, tags, rename, archive, Fork and Hand off (start a new chat seeded with that session's transcript, in this or any other account — see Context hand-off). The Archived view spans every account too, and Restore puts a session back in the account it came from.
• Managers — MCP servers, agent library + AI-generate, hooks + AI-generate, accounts — same operations as the TUI, with the same diff-approval gate for AI-written files (jobs run server-side, you approve a git-style diff before anything is written).
• Usage banner — one live bar-row per account (session/weekly/model windows with reset times), auto-refreshes every minute, refresh button for an immediate re-fetch.
• Stacked toasts — multiple simultaneous notifications (errors, success, info) stack instead of overwriting; each auto-dismisses after 3.5 seconds. Job failures show the error message rather than a generic "Failed".
• Job cancel — running background jobs (plan generation, memory build, review) show a Cancel button; the cancelled flag is cooperative (checked at loop top, no thread kill).
• Persistent preferences — theme and account selection saved to localStorage, restored across page reloads.
• Editable Plan → Execute — the generated plan appears in a monospace textarea for inline editing before approval; "Re-plan" sends feedback to regenerate; "Per-step approval" gates execution step by step. Every generated plan is auto-saved.
• Skills — a list beside what you picked out of it. The four scopes Claude Code actually loads (personal, project, plugin, built-in) are one scrolling list of rows, each carrying the command you type and how often Claude Code has used it, with a filter and scope chips over the lot; the pane on the right holds the description, the folder and the actions for the row you clicked, and Add a skill until you click one. A row you have picked stays visible through a filter that would otherwise hide it. Agents is the same split — yours beside the ready-made library. See Agents & skills.
• Loops — start a /loop in its own session, watch it fire, end it, and edit the loop.md behind a bare /loop. See Loops.
• Global CLAUDE.md — the first tab of Context: the instructions read in every session on an account, its loop.md sibling and the cross-project conventions worth promoting into it.
• Output styles — by scope, with the active one and the file that pins it named, plus four archeus starters to copy (Terse, Reviewer, Pair, Ship).
• Logs — what archeus itself did and why it failed: its own headless Claude calls, background jobs, the auto-memory scheduler, the failover proxy. Filter by level or by text. Before this page every one of those failures went to a NullHandler. See Logs.
• Skills / Worklog / Review / model-routing panels — Skills manager, worklog toggle + entry history, one-click code review (working diff or staged-only), and OmniRoute free-tier configuration.

Icons are inline Material SVG — no CDN, no emoji.

Themes: 32 palettes, 8 skins, 4 themed worlds

A palette answers "what colours" — 32 of them, authored as real hex (Catppuccin, Tokyo Night, Gruvbox, Rosé Pine, Nord, Solarized and more), applied verbatim to every surface rather than derived from an accent hue. A skin answers "what is this app": corner treatment, border weight, type scale, row density, chassis frame, card entrance and which background scene runs. The two are orthogonal — pick both, or let the palette choose its default skin. Live preview before saving.

Reach for Standard when you want none of that: no chassis frame, the system UI font, hairline borders and the quietest background in the set. It is the plain one, and it mixes with all 32 palettes. (It is not the same as the picker's auto tile, which hands the choice back to the palette and always lands on HUD, Terminal or Brutalist.)

A world commits to a whole look instead: it owns its palette, skin, background scene, icon set, overlay and cursor together, and disables the palette/skin pickers while worn. Four ship — anime, cyber, deck and graph, the last an homage to archeus's own architecture graph.

The background is a single full-viewport scene driven by real state (running jobs, navigation, today's burn), capped and calmed by design, and it stops five ways — hidden tab, blurred window, motion:off, stage:off, or prefers-reduced-motion. Set stage: lite (or off) if your hardware struggles.

Setup

The Edge and browser shells need nothing extra. For the native window, see GUI setup.

The GUI is a local HTTP server on loopback with a per-run secret; the routes it exposes are documented in the API reference.