Conversation
Adds a dependency-free dashboard (public/) served by the existing node:http control API, plus the backend it needs: - account list sourced from .env (scripts/api/envAccounts.js) instead of a hand-maintained copy, so the UI cannot drift from the real accounts - per-account login status derived from the session store's stored cookies (scripts/api/sessionStore.js), so the UI can show which accounts still hold a live Microsoft auth cookie - task scheduling: taskScheduler.js persists tasks to scheduled_tasks.json, taskRunner.js polls and starts them one at a time, and both are wired into server.js behind API_ALLOW_SCHEDULE_WRITE - live log streaming over SSE so the dashboard console mirrors the run logParser: treat ExperimentalWarning as a warning rather than an error. Node prints it on stderr for the SQLite session store, and the old /\bWARN/ pattern missed it, so every run opened with a red herring. scheduled_tasks.json is gitignored: it is per-machine run state, not project config. eslint gains a browser-globals block for public/**. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A Stop from the dashboard previously just killed the process tree, which left Chromium instances to be reaped by the OS and could orphan them. - src/util/Abort.ts: run-wide AbortController plus a registry of live browsers, so an abort can force-close every Chromium it knows about. - Browser.ts registers each launched browser and unregisters on disconnect. - index.ts: SIGINT/SIGTERM now abort first (closing browsers) before flushing webhooks and exiting; the between-accounts delay and the 30-minute Edge browsing task are both interruptible, and the account loop stops rather than starting a fresh login. - processManager.js: writes an __ABORT__ sentinel to the child's stdin because Windows cannot deliver a real SIGTERM to a child process; the existing kill timer still escalates to a tree kill if the bot does not exit in time. Covered by tests/abort.test.mjs. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Toggling Visual Search or 30-Minute Edge Browsing in the Web UI had no effect. The env vars reached the spawned bot correctly, but nothing in the child ever consumed them: applyEnvOverrides() was only called by the Docker entrypoint and the CLI, so loadConfig() read config.json verbatim and both flags stayed false. Add mergeEnvOverrides(), which applies the overrides to an already-parsed config object in memory, and call it from loadConfig() before validateConfig() - so overridden values pass the same schema checks as file values, and config.json is never rewritten. applyEnvOverrides() now preflights and delegates to the same merge, keeping Docker behaviour. Also widen boolean parsing to the shell-style words Load.ts already accepts (1/0, yes/no, on/off, case-insensitive) instead of only true/false, tolerate partial failure so one bad value no longer discards valid overrides, and mask secret values (webhook URLs, bot tokens) in the applied-override log. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
…te task The background Edge browsing activity only posted progress reports to the rewards API; no browser tab was actually browsing. EdgeLiveBrowsing opens a tab in the already-authenticated context and reads for the full session window: scrolls with human-shaped pauses, dwells at the end of an article, and occasionally follows a link instead of jumping back to a feed. Links are restricted to the current host or a bing/microsoft/msn/microsoftedge suffix so a session cannot wander onto arbitrary sites. The session honours the run's AbortSignal at every wait, so a Stop unwinds it instead of hanging, and failures are logged as a warning rather than failing the whole activity. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A-Z walkthrough for users with no coding experience: prerequisites, Node.js setup, starting the dashboard, and using every UI section (accounts, manual login, toggles, running, scheduling, live logs). Documents the non-obvious defaults: API_ALLOW_SCHEDULE_WRITE is off so Schedule Run fails until enabled, Remove only hides an account locally, and Add Account never takes a password. Vietnamese version keeps UI labels, commands, and error strings in English so they match the on-screen text. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Adds GET/PUT /accounts/:id/proxy and a proxy editor to the dashboard, so a proxy can be changed without hand-editing .env and restarting the server. The write path validates against the same rules the bot enforces in src/util/Validator.ts (AccountProxySchema) and src/util/Proxy.ts: four protocols, a bare hostname means http://, port 1-65535, credentials must not sit in the URL, username and password must be set together, and SOCKS cannot carry authentication at all because Patchright does not support it. A proxy the API accepts is therefore always one the next run accepts. Values are stored as ACCOUNT_N_PROXY_* lines in .env, matching where the bot already reads them. Active lines are replaced in place so repeated saves do not accumulate duplicates, and the write goes through a temp file plus rename because OneDrive can lock the target. Commented template lines are never treated as configuration. The stored password is never returned - the read endpoint reports only hasPassword. Leaving the password field blank keeps whatever is on disk, so the UI never has to echo a secret back to the browser. Editing is refused with 409 while a run is active; the change applies to the next run. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
PROXY_GUIDE.vi.md walks someone who has never used a proxy through it: what a proxy actually does, whether they need one at all (a single account does not), which of the four supported types to pick and why SOCKS cannot use a password, how to obtain one, how to enter it in the new editor, and how to tell a dead proxy from a working one before blaming the tool. Also documents the "Use for API requests too" toggle, which is the part most people get wrong. It maps to proxyHttp: off means only the browser goes through the proxy while direct Rewards API calls leave from the local connection, on means both do. The guide recommends leaving it off, since cheap proxies frequently break the API calls and that surfaces as unexplained fetch and timeout errors. Links the two guides to each other - the Web UI guide gains a per-account proxy section, the proxy guide points back at it for install steps. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Adds a section distinguishing a proxy vendor (you get one address plus your own username and password) from a free public proxy list (thousands of IPs shared with everyone). The two look similar enough that buying the wrong one is easy, and the failure mode is quiet: the tool starts fine and the account gets flagged later. Public lists are unusable here - every IP is already in use by hundreds of other people, the addresses die within hours, and there is no credential to control or revoke. Some are deliberately seeded to harvest traffic, which matters when the traffic is a Microsoft login. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.