Cursor Agents · Model Routing
Desktop dashboard that shows how a Cursor Agents session routes work across Opus (parent), Fable (advisor), Worker (Sonnet), and Explorer / Researcher (Haiku).
See CHANGELOG.md for release notes.
The main viewport is an agent workshop: five pixel-cast block agents at fixed stations (Opus desk, Fable consult, Sonnet build, Explore / Research shelves). Agents walk paths and carry code crates into a central Merge portal. On ask or delegate, a small inbound route token (mini block sprite) travels along the pulse path into the target station.
- Avatars. Rect body and square eyes. Role is fill color only: no mouths, round faces, or letter initials.
- Chrome. Header Stage | Insights | History picks which drawer sheet owns the lower panel. The workshop stage stays visible.
- Drawers. Session Bill, Activity, Insights, History/Replay, Session Log, and Stations under the stage.
- HUD. Session cost, active role, subagents, effort, and Haiku token chips.
- Demo scripted Play and Live SSE both drive the stage. History replay (play / pause / scrub / speed) uses the same workshop engine. Leaving History stops transport and leaves the last replay frame until Demo or Live runs again.
- Token and session figures are illustrative, not official Cursor billing.
- Optional Web Audio cues are muted by default. Use the header sound toggle (stored in
localStorage); the slider beside it sets volume. Cues are synthesized, not sampled, and mixed through a master gain and a limiter so overlapping cues cannot clip. They are role-tinted, so an Opus crate and a Haiku crate do not sound identical. Alert snooze also silences workshop audio and desktop toasts until you un-snooze, and stays in effect across an app restart.
Double-click repo-root AgentWorkshop.exe (from npm run pack or npm run build via scripts/copy-exe.js; gitignored *.exe), or run npm start while developing. Window title: Cursor Agents · Model Routing. The live bridge runs in-process in Electron. No CMD window and no separate node.exe console.
A second launch focuses the existing window (single-instance lock) and does not bind another bridge on the same port.
Requires Node 24 or newer.
Windows:
cd /d path\to\agent-workshop
npm install
npm startmacOS / Linux:
cd path/to/agent-workshop
npm install
npm start- Opens the Electron window only (
electron .). - Bridge listens on
http://127.0.0.1:3847/in the same process. - Optional port override (default
3847):
set CURSOR_BRIDGE_PORT=3848
npm startCURSOR_BRIDGE_PORT=3848 npm startIn Live mode with the Cursor source selected, a project dropdown lists Cursor projects that have ~/.cursor/projects/<id>/agent-transcripts (newest activity first by default). With the Claude Code source the dropdown lists Claude Code projects instead, and the bridge selects the newest one itself when nothing is selected yet (at start, or when you switch to Claude Code), so the picker shows the project being read and tailing starts without a click. A selection you made is kept, also across a switch to Cursor and back.
Choosing a project:
POSTs/api/selectwithX-Cursor-Bridge: 1and switches the bridge transcripts folder.- Cursor only: opens the History sheet and reloads Insights onto the latest non-empty transcript session for that project (
latestSessionIdfrom the bridge). - Cursor only: remounts the History replay transport. Both sources restart Live SSE (
resetresyncs the workshop; it does not remount History replay or move the scrubber by itself).
With the Claude Code source a pick stays on the current view (History opens from its header tab) and reloads the session list for the new project.
While Live is connected, loading a Cursor session into History (project change or session picker) paints the replay frame on the stage but does not pause Live SSE until you scrub the timeline or press Play. Scrub or Play sets replay ownership so Live events no longer overwrite the stage until you leave History or run Demo/Live again. With the Claude Code source, choosing a session in the History picker changes nothing on the stage, Bill or log; only a scrub or Play takes the stage. Leaving History then puts that Live state back (Bill, log, stage and cards, and today's Runs, tokens and fastest) and applies, quietly, the Live rows that arrived meanwhile, so no crate or sound cue, ship celebration or fastest timing plays for them; a Live finish or error still alerts. Only a replay that queues more than 20,000 Live rows falls back to rebuilding Live from the bridge's recent backlog.
Live mode has a Claude Code | Cursor switch in the header. Claude Code is the default source. Your choice is stored in the browser (ca.source.v1); with no stored choice, or an invalid one, the page asks for Claude Code, and a stored choice (including Cursor) still wins. A bridge starts on Claude Code too, so the two agree on the first connect and the page sends nothing. The dashboard sends it to the bridge with POST /api/source ({"source":"claude-code"} or {"source":"cursor"}, with the X-Cursor-Bridge: 1 header and the same-origin check). A HUD chip shows which source the stage is reading.
What Claude Code mode reads, one project at a time from the picker:
- Windows:
~/.claude/projectson the local drive. - WSL:
/home/<user>/.claude/projectsin each distro that is running. The bridge askswsl.exe -l --running(from%SystemRoot%\System32) and reads each running distro through its\\wsl.localhostor\\wsl$share. A stopped distro is skipped and reported in the project list notes (for example "Ubuntu skipped: not running"), so a missing folder never looks like an empty one. - Labels are path-free. A project shows its own folder name. A home folder, or a project with no working directory, shows a short label instead. Project ids and those labels are HMAC-SHA256 over a random 16-byte salt, made once per install and kept in a config folder of its own (
%LOCALAPPDATA%\cursor-agents-model-routing-configon Windows,~/Library/Application Support/cursor-agents-model-routing-configon macOS,$XDG_CONFIG_HOME/cursor-agents-model-routing-configor~/.config/cursor-agents-model-routing-configon Linux; a relativeXDG_CONFIG_HOMEis ignored), not beside the event log, so a zipped log folder does not carry it. A salt an earlier version left in the event-log folder is moved to the config folder on the next start, so ids and labels stay the same. If the old folder cannot be checked or that old file cannot be read at that moment (in use, a disk error), the run uses a temporary salt (ids and labels change until the next start, with a "Labels change after restart" note), nothing is written to the config folder and the old file is left alone, so the next start moves the real one; if the config folder already holds a valid salt, that salt is used instead, with no note. On Linux and macOS the folder is private (0700) and the file 0600; on Windows the folder relies on your profile's folder permissions.
What you see:
- Claude Code rows drive the existing stage and station cards: the parent session, and a delegation by its
subagent_type(Planner, Researcher, Worker or Reviewer; other types take a role from their model family: opus → Planner, sonnet → Worker, haiku → Researcher, anything else → Worker). In Live on Claude Code the stations are labelled Lead, Planner, Worker, Researcher and Reviewer (Cursor and Demo keep the Cursor labels), and each role lights its station: Lead → first station, Planner → second, Worker → third, Researcher → fourth, Reviewer → fifth. The Bill table keeps its own Cursor-style row names and rates. - Model aliases (
opus,sonnet,haiku,inherit) are resolved to full model ids at ingest, andinherittakes the Lead's model; an unknown alias stays unset. These ids go to the role-profile model table, which shows on the stations while the Claude Code profile is active (Live on Claude Code). - Bill: tokens come from each transcript line's
usage, using the highest count seen for each message id, and are priced by model family with the rates indashboard/bill-math.js(the resolved full model id does not reach the Bill). The Bill view itself is unchanged. Cache tokens are not billed. - History shows older sessions, replay only. The History tab lists the selected project's sessions from the last 14 days (the 20 newest, labelled with local time, event count once opened and duration). Choosing one loads it into the replay transport; Play or scrub drives the stage, and Live keeps updating the stage until you do (leaving History after that restores Live). A session cut at the size limits says so. The list shows an event count only once a session is opened, and it includes the session Live is tailing right now. Insights (forecast, suggestions) does not read these sessions. A project pick stays on the current view.
GET /api/historyreturns no events with ahistory-offnote andPOST /api/history/clearreturns 409; Claude Code events never reach the Cursor event log. - Replay, then alerts. When a session is first read, its catch-up replays silently: rows fill the stage, but nothing alerts, counts toward today's stats, or plays the ship celebration. Once the parent session has been read to its end, finish and error rows alert as they arrive. A newer session clears the stage and replays from its start.
- Switching back to Cursor restores the Cursor project and watch state.
Windows portable .exe (primary launch, silent GUI):
npm run packBuilds the portable single-file exe and copies it to repo-root AgentWorkshop.exe (via scripts/copy-exe.js). Double-click that root file. No install, no sibling DLLs.
Full pass with the NSIS installer:
npm run buildmacOS / Linux:
npm run pack:mac # dmg + zip
npm run pack:linux # AppImage + deb
npm run build # current platform, all configured targetsArtifacts land in dist/. The repo-root AgentWorkshop.exe copy is Windows-only; on other platforms scripts/copy-exe.js logs a skip instead of failing. macOS builds are unsigned (no Developer ID), so Gatekeeper needs a right-click → Open on first launch.
| File | Use |
|---|---|
AgentWorkshop.exe (repo root) |
Primary launch: portable GUI, no install |
dist\AgentWorkshop.exe |
Same portable artifact before/after copy |
Cursor Agents Model Routing Setup x.x.x.exe |
Optional NSIS installer (npm run pack:win) |
Packaged apps use the Windows GUI subsystem: no console window on launch. There is no .cmd wrapper.
Do not copy dist\win-unpacked\*.exe to the repo root alone. That naked Electron binary needs ffmpeg.dll beside it. Use npm run pack:dir only to inspect an unpacked tree locally.
Git already ignores dist/, *.exe, and node_modules/. Source of truth is GitHub + npm install / npm run pack. Safe to delete locally anytime:
- Entire
dist\folder (unpacked trees, portable duplicates, NSIS setup, blockmaps) - Stale root
AgentWorkshop.exe/ legacy*Portable*.exe(recreate withnpm run pack) - Electron/builder caches under
%LOCALAPPDATA%\electron*/%LOCALAPPDATA%\electron-builder*if you are reclaiming a lot of space (next pack re-downloads)
Keep node_modules\ unless you are ready for a full npm ci / npm install (large reinstall). Never delete .git or source files.
| Path | Role |
|---|---|
model-routing-dashboard.html |
Markup shell (Demo + Live UI) |
dashboard/styles.css |
Dashboard styles |
dashboard/cards.css |
Station card / avatar / ship-celebration styles |
dashboard/insights.css |
Insights / History drawer styles |
dashboard/tips.css |
TipLayer floating tooltip styles |
dashboard/alerts.css |
Alert center bell / popover styles |
dashboard/app.js |
Orchestrator: Demo playbook, Live SSE client, mode switching, project picker, jump-to-agent, ship celebration, __cursorAgentsDashTest |
dashboard/activity.js |
Activity drawer, Session Log, and their empty states; owns its row list, log clock, and role filter |
dashboard/view-mode.js |
Header view switching (Stage/Insights/History), the Panels toggle, and the drawer-sheet focus policy |
dashboard/projects.js |
Live project picker: list projects, switch the bridge scope, hand it to Insights + History |
dashboard/presence.js |
Live presence (hint, banner, coach, empty-log copy), the bridge sync-age chip, and its poll |
dashboard/bill-math.js |
Pure bill/usage math: rates, token and money formatting, model normalization, summary CSV · no DOM |
dashboard/bill.js |
Session Bill: usage counters and their DOM, savings comparison, HUD totals; owns the demo daily baseline |
dashboard/demo-playbook.js |
The scripted Demo narrative: one closure per story beat; app.js schedules them |
dashboard/safe-html.js |
The single HTML escaper (window.SafeHtml); every other dashboard module aliases it · loads first |
dashboard/sound.js |
Workshop audio cues (Web Audio, synthesized, no assets): master gain + limiter, volume, role-tinted timbres · muted by default |
dashboard/insights.js |
Insights forecast, cost-ranked suggestions, History/Replay UI |
dashboard/claude-history.js |
Claude Code History: older sessions picker (last 14 days) feeding the replay transport |
dashboard/claude-replay-guard.js |
Keeps the Live state a Claude Code replay covers, and queues the Live rows that arrive meanwhile |
dashboard/replay-stage.js |
Replay transport to workshop stage |
dashboard/cards.js |
Station cards, avatars, daily stats, privacy meter, ship celebration |
dashboard/tips.js |
Unified TipLayer (data-tip / data-tip-detail) |
dashboard/alerts.js |
In-app alert center (SSE alert + GET /api/alerts) |
live-bridge.js |
Headless/Electron bridge (HTML + /dashboard/* + APIs) |
electron/main.js |
Desktop window; starts bridge in-process |
test/demo-smoke.test.js |
Demo-mode Playwright smoke test |
test/cards-smoke.test.js |
Station cards / avatars / Live timing smoke |
test/insights-smoke.test.js |
Insights / suggestions / jump-to-agent smoke |
No server required:
- Open
model-routing-dashboard.htmlin a browser (file://). Relative links loaddashboard/styles.cssanddashboard/app.js. - Leave Demo selected and click Play. Watch the workshop narrative: asks, then explore/research, then worker edit, then advisor, then ship.
- Click a character to filter Activity. Click the Merge portal to open Activity. Use Panels for Bill, Log, and Insights (Stage view). Switch Insights or History for forecast or replay. On History, play / pause / scrub / speed drives the workshop stage. Optional sound: header mute toggle.
npm install
npm run lint
npm run format:check
npm test
npm run build:checkOr npm run ci for the same sequence. npm test needs Playwright Chromium once: npx playwright install chromium.
For browser-only Live mode without Electron:
npm run bridgeor:
node live-bridge.jsThen open http://127.0.0.1:3847/ and click Live.
The bridge listens on loopback only and rejects non-localhost Host / foreign Origin (including Origin: null; no Access-Control-Allow-Origin: *). Open the dashboard from the bridge URL or Electron, not from another website. Demo via file:// does not need the bridge. Clicking Live from file:// points at the client default http://127.0.0.1:3847/ (file pages cannot read CURSOR_BRIDGE_PORT / --port).
Options:
node live-bridge.js --port 3847
node live-bridge.js --dir "C:\Users\<you>\.cursor\projects\<project>\agent-transcripts"Naming a Cursor folder (--dir, or the CURSOR_TRANSCRIPTS environment variable) is an explicit ask for Cursor, so the bridge then starts on the Cursor source instead of Claude Code. The named folder wins for that run: the page shows Cursor on connect (also with no stored choice or a stored Claude Code choice) and does not switch the bridge or change the stored choice. Clicking Claude Code in the header still switches the bridge and stores that choice (known issue 33). CURSOR_BRIDGE_PORT / --port also apply here. start-live.cmd is a legacy headless helper only. It is not the primary desktop launch path.
| Source | Use |
|---|---|
~/.cursor/projects/*/agent-transcripts/**/*.jsonl |
Projects listed in the picker; newest selected by default |
GET /api/projects |
Project list + current selection |
POST /api/select + X-Cursor-Bridge: 1 |
Switch transcripts folder + reset watch state (JSON {id}; Sec-Fetch-Site) |
SSE GET /api/stream |
Events + reset when the project changes; also event: alert |
GET /api/alerts |
Last ~50 in-memory alerts ({ id, ts, name, status, agentId? }) |
GET /api/events, GET /api/status |
Debug / backlog |
GET /api/history?since=&session= |
Persisted history events (JSONL-backed; localhost Host/Origin guards) |
POST /api/history/clear + X-Cursor-Bridge: 1 |
Wipe local history ({"ok":true}) |
In Cursor mode, roles are inferred from Task / CallDynamicTool subagent_type and subagent prompt text (advisor, worker, explore maps to explorer, researcher). Unknown types map to parent/generic. A delegation's arguments.model is read too, so a station can name the model the work actually routed to.
In Claude Code mode the same routes apply, with POST /api/source switching the source. GET /api/status and GET /api/events then return counts and path-free rows only. The Claude Code tailing and wiring are described in Claude Code source.
The Live UI tracks a single presence state for the header hint, stage banner, and empty copy:
| State | What you see |
|---|---|
file-origin |
Coach: open Electron or http://127.0.0.1:<port>/ (Live from file:// is unsupported) |
connecting |
Soft connecting banner |
bridge-down |
Persistent banner + Reconnect (manual; no auto-reconnect loop) |
no-projects |
Coach when the picker has no projects for the source in use (Claude Code or Cursor) |
quiet-healthy |
Connected with no recent events. Idle workshop, not an error. |
streaming |
Active traffic + project chip |
First time you enter Live, a dismissible coach sheet explains the project picker, what Live sees, and the estimate disclaimer (localStorage ca.liveOnboard.v1). Demo is never blocked.
The Insights sheet shows a session forecast, cost-ranked Suggestions, and TipLayer help text. Suggestions sort by estimated dollar savings, then name. Clicking a row or an alert with a mappable agent runs jumpToAgent: it selects that station's block agent on the stage, opens Activity, and briefly highlights the card. The Stations drawer lists agent cards with pixel-cast block avatars, daily stats, and a privacy meter. None of this changes Cursor routing. It only navigates the dashboard.
Chrome, stage stations, HUD chips, station cards, Insights, History/Replay, and the alert center share one TipLayer hover/focus tip (data-tip / data-tip-detail, ~300ms delay). Prefer that over bare title= on primary controls. Ship / merge plays a short portal celebration and marks avatars done (respects prefers-reduced-motion).
History and GET /api/history show events for sessions that exist under the currently selected project's agent-transcripts tree. The bridge filters persisted history to those session IDs. If the folder is empty or has no JSONL yet, History shows no events for that project.
Replay transport lives in the History sheet and shares Insights replay UI. In Live mode, see Live project picker above for when SSE keeps streaming versus when replay takes the stage.
The bridge also keeps a local event log (JSONL) under the OS app-data profile, not in the git repo:
| Platform | Default path |
|---|---|
| Windows | %APPDATA%\cursor-agents-model-routing\event-log.jsonl |
| macOS | ~/Library/Application Support/cursor-agents-model-routing/event-log.jsonl |
| Linux | ~/.local/share/cursor-agents-model-routing/event-log.jsonl |
Overrides: CURSOR_EVENT_LOG (file) or CURSOR_EVENT_LOG_DIR (directory). Entries older than 30 days are pruned when the bridge starts. Values that look like API keys, tokens, passwords, or bearer secrets are redacted before write.
History event contract (stable):
{
"ts": "2026-01-01T00:00:00.000Z",
"sessionId": "string",
"agentId": "string",
"parentId": null,
"name": "string",
"model": "string",
"modelRaw": "string",
"type": "spawn",
"inputTokens": 0,
"outputTokens": 0,
"files": []
}type is one of spawn | turn | done | error. model is the billable rate bucket; modelRaw is the Cursor slug a delegation pinned (empty when the child inherited the parent's route), and is display-only.
Terminal rows come from the transcript's turn_ended line, which is the only signal Cursor writes. When an agent finishes, errors, or stalls (no activity for about 2 minutes), the bridge:
- Broadcasts SSE
event: alertwith{ id, ts, name, status, agentId? }and keeps the last ~50 forGET /api/alerts. - In Electron, also fires a desktop Notification (name + status only).
Which terminals alert. A subagent is short-lived, so its finish or error always alerts. The parent ends a turn on every user message, so a parent success stays quiet — otherwise a long session would stream one toast per turn and each would claim the session had finished. A parent error still alerts, except when the transcript says the user stopped the turn (user_stopped_generation, composer_abort_controller_aborted). Repeat terminals for one agent are deduplicated. The error's own text — including a quota notice like Other Models usage limit reached — rides the event feed, never the alert payload, which is kept to name and status so it cannot leak a path or a prompt.
The in-app alert center (header bell) lists the same alerts: mark read, snooze, clear, click to jump when a role/agent is mappable. Mappable roles show a small block-agent avatar beside the status (same cast as Stations, never letter initials). Alerts show only agent name and status, never prompt text or file paths. Backlog/SSE replay quiet gates skip spam the same way card timing does.
- Near-real, not hook-level realtime (no Cursor hooks required for this slice).
- SESSION BILL is transcript-estimated tokens times published rates per 1M input/output: Opus $4/$20, Sonnet $2/$10, Haiku $0.10/$0.50, Fable $8/$40. Real transcripts carry no
usageblock, so every figure is estimated from character counts — treat it as an estimate of magnitude, not a count. Estimates are illustrative, not official Cursor invoices. - A delegation that names a model outside the four published families (Composer, Grok, GPT, Gemini, DeepSeek) is labelled by name but still priced at the delegated role's rate, because the bill table only covers four families.
- Timestamps are ingest-time. Real transcripts have no timestamp field, so History replay reflects when the bridge read a line, not when the session produced it.
- ACTIVITY preview lists reads and edits from the transcript (path + short detail). It does not include full file contents.
- Only sees chats that write into the selected project's transcripts folder.
- A real delegation carries no child agent id, so the parent-side
spawnrow cannot be keyed to the child. The subagent's own transcript supplies its spawn row, and the parent's delegation is recorded as aturnrow carrying the model. - History API responses omit events whose
sessionIdis not present under the current transcripts tree. - History records only bytes the bridge saw arrive. A transcript that already existed when the bridge started is warmed from the tail, not back-filled.
- Hooks, statusline, and Cursor SDK are separate integration paths. This dashboard reads transcripts on disk.
- Reads only the newest session of the selected project, plus that session's subagents. One project at a time.
- Files are polled about every 900 ms: near-real, not hook-level realtime.
- The subagent list holds about 1000 children per session (the first 2000 directory entries are read, and each child has a
.meta.jsonbeside it). - A finish in the first ~4.5 seconds of a brand-new session never alerts: the switch waits 5 polls, then the catch-up replays silently.
- Replay rows never count toward today's stats. A subagent that ends
failed,errororkilledplays the error sound once when it arrives live (not during the catch-up);stoppedandcancelledstay silent. - Only tokens from
usageare billed; cache tokens are not. The Bill view is the Cursor one, unchanged. - The Cursor event log is not used in this mode. History lists older sessions for replay only (last 14 days, 20 newest); an event count shows only once a session is opened, and the list includes the session Live is tailing right now.
- On a first start, two processes can race to create the salt on a disk without hard links (FAT, exFAT, most network shares). See
docs/known-issues.md(items 27 and 36).
The bridge reads Cursor's on-disk transcripts. This list is what the code actually reads, checked against real transcripts on disk rather than assumed. It was rewritten after an audit found the previous list described a shape Cursor does not write.
- Path layout.
~/.cursor/projects/<project-id>/agent-transcripts/<session-uuid>/<session-uuid>.jsonl, with subagent transcripts at<session-uuid>/subagents/<subagent-uuid>.jsonl. Override the root withCURSOR_PROJECTS_ROOT. - Two line shapes. One JSON object per line, of exactly two kinds:
- Conversation:
{"role":"user"|"assistant","message":{"content":[…]}}. Content parts aretext,tool_use, ortool_result. - Terminal:
{"type":"turn_ended","status":"success"|"error","error"?:string}. Cursor ends every turn with one, in parent and subagent files alike.
- Conversation:
- Timestamps. There is no timestamp field. Time is prose inside the user text (
<timestamp>Thursday, Oct 15, 2026, 9:12 AM (UTC-4)</timestamp>) and is not read. Every event is stamped when the bridge ingests it, so History replay shows ingest order and ingest time, not the session's real clock. - Roles.
CallDynamicToolwithinput.toolName === "Task"carriesinput.arguments.subagent_type, which maps to advisor, worker, explorer, or researcher. A subagent transcript infers its role from prompt text. Anything unrecognized renders on the worker card with anunrecognized rolemarker. - Model.
input.arguments.modelon a delegation names the model, and the station shows it (claude-haiku-5-5-thinking-high→ Haiku). Only about a third of delegations name one; the rest render with the role's default label.inherit,fast,auto, anddefaultare route words, not models, and render as no label. A model family with no published rate in this build (Composer, Grok, GPT, Gemini, DeepSeek) is named but still estimated at the delegated role's rate, because the bill only prices four families. - Terminal signals. Read from the
turn_endedline, never from prose words or assistant tags. A transcript that ends without one stays un-terminated. - Usage. There is no
usageblock in any real transcript. Every token figure is estimated from character counts at roughly four characters per token, so Session Bill estimates magnitude rather than restating a real count. - Drift. A line that does not parse, parses to a shape this build does not know, or ends a turn with a status this build does not recognize, is counted and surfaced instead of dropped.
GET /api/statusexposesdrift: { unparsed, unknownShape, unknownStatus, lastAt, lastSample }, and the first one of a session also lands in the event feed. A turn with an unfamiliar status is still recorded, but never guessed as done or error. Non-zero means the on-disk format moved.
Last verified against real transcripts on 2026-10-16. Cursor controls this format and can change it; there is no version negotiation. If the dashboard stops seeing sessions, check this list first, then drift in /api/status.
Unknown roles are visible. A subagent_type the bridge does not recognize is marked unrecognized role on its delegation line in the Session Log and Activity drawer. The type renders as a worker card either way, so the marker is the signal that Cursor introduced a role this build does not know. Telemetry is local only; nothing is sent anywhere.
Contributions are welcome. npm run ci is the gate; see CONTRIBUTING.md.
The bridge binds loopback only and has no authentication beyond that — read
SECURITY.md for the threat model, the guards, and the accepted risks
before running it on a shared machine. Report vulnerabilities privately, not in an issue.
