Safe mode: start with tools, schedules and channels off

Aperta
#236 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
24/100
Tipo di issue
Funzionalità
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
rust, tauri

Direzione di ricerca

Read the Serve command in apps/cli/src/main.rs and trace how daemon startup wires the scheduler and channels; then read docs/guide/troubleshooting.md. The proposal calls for gating startup services, tools, schedules, and run resumption, plus desktop UI, status reporting, translations, and tests. First settle the exact scope in the PR description; done means the acceptance criteria pass without changing stored settings.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

area: daemon area: desktop-ui enhancement help wanted

size: M · first-timer: no

Summary

A safe mode that starts Praxi with everything that acts turned off, so a user can open the app, look at their bots and chats, and fix a problem without anything running by itself.

Why it matters

When something goes wrong (a schedule that starts a failing run on every launch, an MCP server that hangs at startup, a skill that misbehaves, a bot that keeps controlling the screen), the app is hard to use precisely because it starts working immediately. People need a way to get in, read, and change settings first.

Current state

  • There is no single "do nothing" switch. bot doctor checks the installation (crates/daemon-api/src/doctor.rs), each bot's tool set can be filtered (see the scenario evals/scenarios/11-tool-filter-hides-tool.yaml), and schedules, MCP servers, skills and channels are managed one by one (bot schedules, bot mcp, bot skills, bot channels).
  • At start the daemon (bot serve, apps/cli/src/main.rs) brings up the scheduler, the Telegram bridge, configured MCP servers and, on a host, peer networking; the desktop app starts the daemon and connects to it.
  • Interrupted runs are resumed after a restart (Serve docs: live runs are paused on shutdown and resumed after the next start).

Proposal and suggested approach

Settle the exact scope in the PR description, then implement:

  1. Entry points: bot serve --safe (also BOT_SAFE_MODE=1) and a "Start in safe mode" choice in the desktop app's recovery or startup screen. It is decided only at process start by the local owner: no tool, route, bot or paired device can switch it on or off at runtime.
  2. What it does: the model can still answer in chat, but every tool is hidden and refused; schedules and workflows do not fire (and are not deleted or changed); the Telegram bridge, MCP servers, skills, background memory work and peer or host networking do not start; interrupted runs are not resumed automatically (they stay paused and visible).
  3. UI: a persistent banner "Safe mode: tools, schedules and channels are off", with a clear way to restart normally.
  4. Diagnostics: bot doctor and the status route report that safe mode is on.
  5. Tests: with a fake provider, a bot cannot call any tool in safe mode, a due schedule does not run, and the normal start is unchanged; a unit test that no API route changes the mode.

Acceptance criteria

  • Starting in safe mode never changes stored bots, chats, schedules or settings.
  • Nothing that touches files, the shell, the browser, the desktop, the network (other than the model call) or other apps runs in safe mode.
  • The daemon and the UI both show the mode, in English and Turkish.
  • cargo test -p cli -p daemon-api -p agent-core, check:i18n, UI tests and clippy pass.

Out of scope

A security sandbox against malware (safe mode is a debugging aid, as docs/guide/security.md says about approvals generally), per-bot safe mode, and automatic entry into safe mode after a crash (a possible follow-up).

How to get started

Read the Serve command in apps/cli/src/main.rs and how the daemon wires the scheduler and channels at startup, then docs/guide/troubleshooting.md. Comment here with the list of subsystems you would gate.

Lingua principale
Rust
Stelle
0
Fork
1
Merge medio
6h 54m
PR unite (30g)
152

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di emreocell/Praxi-Bot

Tutte le issue di emreocell/Praxi-Bot

Issue simili

Altre issue su Rust

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.