OOPSpam
Identity & SecurityConnect AI agents to OOPSpam, a spam detection API for forms, comments, and signups. Agents score submissions for spam, check a domain against DNS blocklists, and report misdetections back for retraining.
What This Integration Enables
Plenty of services will tell you whether something is spam. OOPSpam is one of the few that ships the correction path as a first class operation, and that is the reason to choose it for a form pipeline you intend to keep tuning. Check Spam analyzes a submission and returns a Score from 0 to 6 along with a Details breakdown showing which signals contributed. You supply at least one of Content, Sender IP, or Email, and more of them produce a better answer. The optional checks cover blocking disposable and temporary email addresses, blocking VPN, proxy, and Tor IPs, blocking data center IPs, a minimum length check, URL friendly scoring, restricting to allowed languages, and allowing or blocking sender countries by IP geolocation. A score of 3 or higher is generally treated as spam. Supplying a Business Context switches the call to contextual detection, which replaces the standard content check and is what separates a real enquiry about your product from a fluent, well written message about somebody else's.
Two constraints shape how you wire it. The country, VPN, and data center checks all require a Sender IP, so a form that does not capture the submitter's address silently loses those options rather than erroring. And dashboard logging requires a Source identifier, which is what lets you tell your contact form's traffic apart from your comment form's when you go looking at what the filter has been doing. Check Domain Reputation is a separate read that checks a domain against multiple DNS blocklist providers and returns whether it is currently Blocked plus the Blocker list naming which providers list it, which is useful both for scoring inbound senders and for checking that your own sending domain has not been listed. Report Misdetection is the only write in the connector, and it is the only call whose effect outlives the submission that triggered it. There are no triggers here: nothing in OOPSpam starts a FlowRunner flow and there is no event subscription, so the check runs at the moment your form, comment, or signup handler receives the submission.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Scoring Inbound Demo Requests
A demo request form on the site submits through Typeform or a native handler. The agent runs Check Spam with the content, the sender IP, and the email address, with disposable email blocking and data center IP blocking enabled and a Business Context describing what your product actually is, so a polished message selling SEO services scores as what it is rather than as a fluent enquiry. Submissions that score below the threshold create the lead in HubSpot and post to the sales channel in Slack. Everything else goes to a held folder with the score and the Details breakdown attached, because a lead you filtered is still a lead until somebody has looked at it.
Comment Moderation With an Audit Trail
A community or blog receives comments. The agent runs Check Spam on each one with a Source identifier set so the dashboard separates this stream from the contact form, and with allowed languages configured to match the audience. Comments scoring at or above the threshold are held rather than published, and the moderator sees the score and the contributing signals alongside the text. Nothing is deleted by the agent. The record of what was filtered and why is what lets the team notice a month later that the language restriction is catching a legitimate segment of their readers.
Checking a Domain Before It Costs You
Two uses, one action. On the inbound side, when a submission sits near the boundary, the agent runs Check Domain Reputation on the sender's email domain and adds the Blocked status and the Blocker provider list to the case, which often resolves an ambiguous score. On the outbound side, a scheduled flow runs the same action against your own sending domains and alerts the marketing owner the moment one appears on a blocklist provider, which is usually discovered through falling open rates weeks after it happened. The same read serves both, and the second one is the one people forget to build.
Human-in-Loop Highlight
Check Spam scores one submission. Report Misdetection changes what happens to all of them. It is the only write in this connector, it goes to OOPSpam rather than to a system you control, and setting Should Be Spam to false on something that was in fact spam teaches your account's detection to let that pattern through from then on, for every form and every submitter, with no undo operation anywhere on this surface. That makes it a genuinely poor fit for automatic execution, and it is exactly the call an agent is most tempted to make: a submission is held, a salesperson says it looks fine, and reporting it back as legitimate is a single obvious step. So the agent does everything up to that step and stops. It holds the submission, records the score, the Details breakdown, and the Check Domain Reputation result, and posts to the marketing owner: "Submission from [email] scored [n] and was held. Contributing signals: [list]. Domain reputation: [Blocked or clear, with providers]. Release to the queue only, or release and report it to OOPSpam as legitimate, which will affect future scoring on this account?" Releasing one message and retraining the filter are presented as two different decisions, because they are. Scoring, holding, routing, and alerting all run without a person. The correction that outlives the message waits for one. That is human-in-the-loop placed on effect rather than on volume.
Agent Capabilities
3 actionsSpam Detection
2- Check Spam Analyzes a form, comment, or signup submission for spam and returns a Score from 0 to 6 along with a Details breakdown of which signals contributed. Combine any of Content, Sender IP, and Email, of which at least one must be provided. Optional checks include blocking disposable and temporary emails, VPN, proxy and Tor IPs, and data center IPs, a minimum length check, URL friendly scoring, restricting to allowed languages, and allowing or blocking sender countries by IP geolocation. A score of 3 or higher is generally treated as spam. Providing a Business Context switches to contextual detection in place of the standard content check. Country and VPN or data center checks require a Sender IP, and dashboard logging requires a Source identifier.
- Report Misdetection Reports a spam detection result that was classified incorrectly so future detections improve. Resubmit the same content, sender IP, or email that were checked and set Should Be Spam to true to mark the submission as spam or false to mark it as legitimate. This changes future scoring for the account rather than only the submission in hand, and it has no reverse operation, so it is routed for confirmation before it runs.
Reputation
1- Check Domain Reputation Checks a domain against multiple DNS blocklist providers and returns whether it is currently Blocked and which providers list it. Used to resolve borderline inbound scores, and on a schedule against your own sending domains to catch a listing before it shows up as falling deliverability.
Frequently Asked Questions
What can FlowRunner do with OOPSpam?
FlowRunner agents can run Check Spam, Report Misdetection, and Check Domain Reputation in OOPSpam.
Does connecting OOPSpam to FlowRunner require OAuth?
No. OOPSpam connects to FlowRunner with an API key, no OAuth flow required.
Can OOPSpam trigger a FlowRunner workflow automatically?
OOPSpam doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with OOPSpam
$100 in credits. No card required. Connect in minutes.