Brand Mentions Monitor: Hacker News, Bluesky, GitHub, Lemmy
Pricing
from $0.40 / 1,000 mentions
Brand Mentions Monitor: Hacker News, Bluesky, GitHub, Lemmy
Find public mentions of your brand, product or keyword on Hacker News, Bluesky, Lemmy, Mastodon, GitHub issues and DEV in one run: text, link, author handle, time, likes and replies. Monitor mode returns only new mentions. No API keys. $0.40 per 1,000 mentions.
Pricing
from $0.40 / 1,000 mentions
Rating
0.0
(0)
Developer
ActorForge
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Categories
Share
Find public mentions of your brand, product or keyword on Hacker News, Bluesky, Lemmy, Mastodon, GitHub issues and DEV in one run: text, link, author handle, time, likes and replies. Monitor mode returns only new mentions. No API keys. $0.40 per 1,000 mentions.
Monitor public mentions of your brand, product, library or any keyword across the open developer and tech communities, in one run and one table. Give it keywords and it searches Hacker News (stories and comments), Bluesky (posts and replies), Lemmy (posts and comments on large federated instances), Mastodon (public hashtag timelines), GitHub (issues and pull requests) and DEV / dev.to (articles by tag), then returns one normalised row per mention: source, link, title and text snippet, the public author handle, time posted, likes, replies, reposts and score, where it was posted (HN thread, Lemmy community, GitHub repo, hashtag) and which keyword matched. Filter by time window (last hour to last year), exclude noise words, drop comments, pick a Bluesky language or author. Switch on monitor mode for scheduled runs and each run returns only mentions you haven't seen yet, which makes it a cheap social listening and brand monitoring feed for devrel teams, open-source maintainers, developer-tool startups, PR and competitor tracking, alerts to Slack via Make, n8n or Zapier, and data for AI agents. Every source is an official or documented public API used keylessly (an optional free GitHub token raises GitHub's limit); robots.txt and Crawl-delay are respected on community-run servers. Only public posts and public handles are returned: no follower lists, profiles or emails. $0.40 per 1,000 mentions, with no start fee and no compute bill on top. Runs that find nothing return a free status row instead of failing.
What data does it return?
Every run outputs clean, schema-validated rows like these:
| createdAt | source | contentType | authorHandle | channel | text | matchedKeyword | likeCount |
|---|---|---|---|---|---|---|---|
| 2026-10-23T10:35:42.000Z | hackernews | comment | tgv | France needs to read my paper | > Wouldn't a better solution be to more heavily tax… | rust | |
| 2026-10-23T09:39:24.023Z | bluesky | post | silentstars.dev | 🙌 Community pick iota — Open-source five-language dashboard… | typescript | 3 | |
| 2026-10-23T10:42:16.000Z | github | pull-request | iomete-renovate-bot[bot] | iomete/jupyter-renderer-react | chore(deps): update storybook monorepo to ^10.6.3 | typescript | |
| 2026-10-23T07:29:05.614Z | mastodon | post | @Telemetry_tracker@mastodon.social | #typescript | Custom error classes in JavaScript keep name = "Error" unless… | typescript |
How much does it cost?
Transparent pay-per-event pricing — you only pay for what you get, and the actor always respects your run budget: set a maximum total charge and it stops cleanly at the cap instead of overrunning.
| Event | Price | Charged | Worked example |
|---|---|---|---|
mention | $0.0004 | per mention row (post, comment, story, issue/PR or article) | 500 mentions of your brand = $0.20 |
apify-actor-start | $0.00005 | per run per GB of memory (platform start fee) | One run at 256 MB = $0.00005 |
Input options
| Field | Type | Required | Default | Description |
|---|---|---|---|---|
keywords | array | yes | — | Brand names, product names, libraries or phrases to find, e.g. "supabase", "Next.js" or "vector database". Hacker News, Bluesky, Lemmy and GitHub search full text; Mastodon and DEV have no keyless full-text search, so there the keyword is used as a hashtag/tag ("Next.js" → #nextjs). |
sources | array | no | ["hackernews","bluesky","lemmy","mastodon","github","devto"] | Where to look. Hacker News (stories and comments), Bluesky (posts), Lemmy (posts and comments on large instances), Mastodon (hashtag timelines), GitHub (issues and pull requests), DEV (articles by tag). |
maxResultsPerSource | integer | no | 50 | Stop after this many mentions from each source (shared fairly between keywords, newest first). Each mention is one charged row. |
sinceHours | integer | no | 168 | Only mentions created in the last N hours: 24 = one day, 168 = one week, up to 8,760 (one year). |
excludeKeywords | array | no | — | Drop mentions whose title or text contains any of these words or phrases (whole words, case-insensitive), e.g. "hiring" or a homonym. |
includeComments | boolean | no | true | Also return Hacker News comments, Lemmy comments and Bluesky/Mastodon replies. Off = top-level stories and posts only. |
blueskySort | string | no | "latest" | Latest: newest posts first. Top: Bluesky's own ranking of popular posts in the time window. |
blueskyLanguage | string | no | — | Only Bluesky posts tagged with this language, as an ISO code like en, de, ja or pt. Empty = all languages. |
blueskyAuthor | string | no | — | Only Bluesky posts by this handle, e.g. bsky.app or yourbrand.com (useful to track your own account's posts). |
lemmyInstances | array | no | ["lemmy.world","lemmy.ml","programming.dev","sh.itjust.works"] | Lemmy servers to search (federated, so each sees most public communities; duplicates are merged). Each instance's robots.txt and Crawl-delay are respected: lemmy.world allows one request per minute, so many keywords make runs slower. |
mastodonInstances | array | no | ["mastodon.social","fosstodon.org","hachyderm.io"] | Mastodon servers whose public hashtag timelines are read (each also sees federated posts; duplicates are merged). |
githubToken | string | no | — | A free GitHub personal access token (no scopes needed) raises GitHub's search limit from 10 to 30 requests per minute. Without one, GitHub is searched keyless and skipped with a note if its shared limit is hit. |
newSinceLastRun | boolean | no | false | For scheduled runs: skip mentions already delivered by an earlier run with the same keywords and sources, so you only get (and pay for) new ones. Only hashed IDs are kept, in a key-value store named dev-community-mentions-state in your account. |
Limits & honest notes
- Sources and terms: Hacker News via the Algolia HN Search API (documented by HN, no published rate limit; paced at 2 requests/s). Bluesky via the public AppView API api.bsky.app, whose robots.txt says "Crawling the public parts of the API is allowed" (one request at a time). Lemmy and Mastodon instances via their public APIs, with each server's robots.txt and Crawl-delay respected (lemmy.world asks for one request per minute). GitHub via the REST search API under the GitHub Terms (no selling of personal data, no spam; emails are never collected). DEV via the documented Forem API. Not affiliated with any of these platforms.
- Mastodon and DEV have no keyless full-text search, so there the keyword is matched as a hashtag or tag ("Next.js" → #nextjs, "open ai" → #openai). Hacker News, Bluesky, Lemmy and GitHub search the full text.
- Reddit and X/Twitter are not included: their APIs need paid or commercial agreements.
- Privacy: only public posts, their links and public handles are returned; no follower lists, profile details or emails. Bluesky accounts that opted out of logged-out visibility and Mastodon accounts marked noindex are skipped. Monitor mode stores only hashed IDs of delivered rows. Text is a snippet (up to 1,000 characters), not an archive copy; please respect deletions and the platforms' terms when you store or republish.
- GitHub allows 10 keyless search requests per minute per IP (shared on Apify), so GitHub is paced at one request every 6.5 s and skipped with a note if the limit is hit; add a free personal access token (no scopes) for 30/min. With many keywords, Lemmy's Crawl-delay also makes runs slower (about a minute per extra keyword).
- If one source is down or refuses a request, the run logs it in the summary and continues with the others; it never fails because of one source.
- Results are newest first per source. maxResultsPerSource is shared fairly between keywords.
- Respects your run budget: set a maximum total charge and the run stops cleanly at the cap, with nothing charged twice. A run with no matches returns one free status row explaining why.
Related ActorForge actors
- News Search — the same keywords in worldwide news articles
- RSS Feed Reader — follow specific blogs and changelogs
Changelog
- 2026-10-23 — Initial release: keyword mentions from Hacker News, Bluesky, Lemmy, Mastodon, GitHub issues/PRs and DEV, exclude words, Bluesky language/author filters, monitor mode, free status rows.
Built and maintained by ActorForge. Reliability first: every actor in the portfolio is monitored by an automated nightly self-test, and breakages are typically fixed the same day.