Feed Builder — your own feeds the easy way

You already know the list you want. Every host with this DIRECT seat. Every ads.txt that hashes to this file. Recipe sites on Namecheap that actually publish ads.txt. You build it in a spreadsheet. Next week the set moved. You build it again.

Feed Builder is that list, pointed at the corpus. Paste a hash, a GTM container, a seller ID, a registrar, a registration day. We tell you which pivot it is. Stack a few criteria. You get a live count, a sample, and a saved feed. Set it to daily or weekly and we re-run it for you — membership snapshot, who entered, who left. No cron of your own. No re-export.

Paste first. Don't hunt the catalogue.

Drop GTM-… and we pick tracking ID. Drop a 64-hex digest and we rank the file-hash dimensions. Drop ads.txt and that's “has a latest snapshot.” Drop 2024-03-15 and that's RDAP registered-on. Browse the catalogue when you know the dimension. Most of the time you don't. You have the identifier.

Paste an identifier, pick the pivot, stack chips. Preview count and sample update as you go.

Criteria AND. Operators on the chip — is, is not, glob, regex, starts with, before / after. Add one and the preview moves: live count plus a sample. You see it working before you save.

Keyword list? Hostname glob or regex, plus a cut that matters: has ads.txt, a registrar, a nameserver. You get the cohort. Not every recipe subdomain on earth.

The pivots you already use

Identical file. ads.txt / app-ads.txt / sellers.json sha256. Same bytes, different hosts.
Same partners. Constellation hash, TAG cert pattern, OWNERDOMAIN, MANAGERDOMAIN.
Same seat. Ad system + account, DIRECT seat, seller ID.
Same infra. Registrar, registered-on (that day, or before / after), nameserver, exact NS set.
Same tracker. GTM, GA4, pixels — the families we already score in Ad+Privacy Metrics. Not the MFA grade. A grade would dump the lab corpus into a feed.

Criteria: registered on before 2024-03-15 AND registrar is Namecheap, 268 hostnames match — Registrar plus a registration bound. Count updates as you stack.

Daily or weekly. Built in.

Name the feed. Flip schedule to daily or weekly. That's the job. We re-run the same criteria on a timer, write a membership snapshot, and keep the delta. Members tab is the current set (CSV / JSON). Changes tab is who entered and who left. Runs tab is when it last fired. Manual run is still there when you don't want to wait for the clock.

That's the part the spreadsheet never did: automated updates on the feed itself, not a script you have to remember. Pin the page in a project. Come back to the +/− instead of rebuilding the list.

Feeds list with name, criteria, members, change plus and minus, schedule, and last run — Your feeds. Schedule, last run, who entered and who left. New feed when you need another set.

One host you already suspect? Run Connection report on it. Feed Builder is the other direction: start from a criterion, get the set, keep it current.

Go run it

Tools → Feed Builder (decryptads.com/feeds). New feed at /feeds/new. Preview POST /api/feeds/preview. Classify a paste with POST /api/feeds/classify. MCP preview_feed, create_feed, list_feeds, classify_feed_value.

Feature is available to Researcher+.

Related publishers from a seed host: Connection report. Company name or ads ID: Legal dossier. Risk scoring and agent lookups: analytical features, API, MCP.