# Changelog of Telegram Channel Scraper & Keyword Search - Alerts, API, MCP (`akhil_rajesh/telegram-channel-scraper`) Actor

- **URL**: https://apify.com/akhil_rajesh/telegram-channel-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/akhil_rajesh/telegram-channel-scraper.md

## Changelog

### 2026-11-10

- The README links a new Actor: Indeed Jobs MCP Server & API.

### 2026-11-06

- The README links a new Actor: Pinterest Trends Scraper & API.

### 2026-11-04

- The README links a new Actor: Google Trends Scraper & API.

### 2026-11-03

- The Apify SDK is updated from 4.0.2 to 4.2.0 (apify-client 3.4.0 and crawlee 1.11.0 stay). Runs return the same data. A run still stops exactly at your maximum cost per run and says so: SDK 4.2.0 would let one more result through past a spent limit, and this Actor keeps the earlier behaviour, so nothing past your limit is stored or charged.
- With SDK 4.2.0, a status update that doesn't reach Apify is logged instead of stopping the run.

### 2026-11-02

#### 0.3.3 - Change tracking and alerts hold up however a run ends; private webhook addresses refused in every spelling

- With Only changes, a run that the platform had restarted (after moving it to another server) and that was then aborted just as it saved a channel's history could lose the posts its first attempt had delivered: the abort saved the same history at the same moment, and its copy, written last, left them out. The next run then reported those posts as new again and charged for them. Now one save of a history waits for the other to finish, so the second adds to what the first wrote.
- With Only changes, a restarted run that was aborted before it got back to a channel's history also lost the posts its first attempt had delivered there, with the same result. The abort now saves them too.
- With Only changes, a restarted run that read a channel again only in part (Telegram stopped answering partway) left out of that channel's history the posts its first attempt had delivered on the pages it didn't get back to, new and edited posts alike. The next run reported (and charged) them again. They are now kept.
- A tracked run that stopped on an unexpected error, or because the dataset refused to store results, didn't save the history of the posts it had stored, so the next run reported (and charged) them again. Both now save it first.
- A tracked run that was moved to another server, aborted or stopped by an error after delivering some of the new posts could make the next attempt or run skip the older new posts it hadn't delivered yet: they were recorded as known and never reported. They are now reported.
- Alerts are now posted when a run is aborted (in the time an abort leaves, one try per message), stops on an unexpected error, or fails to store results: the posts it stored are posted, because the next run counts them as known and wouldn't post them. A restarted run still doesn't post a message twice.
- When a channel's history doesn't answer, its posts are returned without a change, as before; alerts no longer post an "alerts are set up" message for it, and no longer say "no changes to post": the status names the channels that weren't compared.
- Alert webhook URLs that point to a private or local address written in a short or numeric form (`https://127.1/`, `https://0x7f.1/`, `https://2130706433/`) were accepted. They are now refused with a clear message before any request is made, like `https://127.0.0.1/`. Public addresses written as numbers still work.
- Library versions are fixed: apify-client 3.4.0 and crawlee 1.11.0 next to the Apify SDK 4.0.2, so a rebuild can't bring in versions the tests haven't run on.
- The README links two new Actors: YouTube Channel Scraper and Contact Details Scraper.
