SlopGuard

What is SlopGuard?

SlopGuard is an AI slop PR/issue guard built as a GitHub App for maintainers. It scores incoming pull requests and issues from 0 to 100, points out why something looks machine-generated, and then either passes it through or quarantines it for review.

The app is designed for the maintenance problem that comes with AI-generated submissions: they often look polished for the first glance, but they still cost time to read, verify, and reject. SlopGuard focuses on triage, not enforcement. It labels likely slop, comments with reasons, and leaves the final call to a person.

Key features

• 0–100 slop score: every contribution gets a numeric score so you can set a threshold for review.
• Reasoned flagging: it does not just say “bad”; it surfaces patterns such as boilerplate phrasing, generic titles, or empty descriptions.
• Provenance tagging: the app records provenance clues, including a prompt fingerprint when available in its demo output.
• Quarantine flow: items at or above your threshold can be labeled and commented on instead of silently accepted.
• Human-in-the-loop controls: maintainers can respond with commands such as /slop approve, /slop reject, or /slop false-positive.
• No auto-close behavior: SlopGuard never closes PRs or issues automatically, which reduces the risk of blocking a real first-time contributor.
• One-click install: the site presents it as a GitHub App install for a repo or organization, with no Action YAML, CI setup, or secrets to manage.
• Source-available and self-hostable: the product page says it is source-available and can be self-hosted for your own use.

How to use it

1. Install the GitHub App on the repository or organization you want to protect.
2. Set the slop threshold with the slider or equivalent threshold control.
3. Let SlopGuard read new PRs and issues as they arrive.
4. Review the score, the reasons, and any provenance trail attached to the contribution.
5. Use maintainer commands such as /slop approve or /slop reject when you are ready to decide.

The demo shows the threshold behaving like a line between passed items and quarantined items, so the workflow is intentionally simple: score first, review second, decide last. That makes the AI slop PR/issue guard useful when your queue is full and you need a quick pass over the likely noise.

Who it is for

SlopGuard is aimed at GitHub maintainers who are seeing more AI-generated PRs and issues than they want to inspect one by one. It is especially relevant if:

• your project receives a steady stream of low-signal submissions;
• you want to reduce triage time without blocking genuine newcomers;
• you need a clear audit trail for why something was flagged;
• you prefer a quarantine-and-review approach over automatic rejection.

The app’s positioning is narrow and practical. It is not a general moderation platform, and it is not trying to replace maintainer judgment. It is an AI slop PR/issue guard for teams that want help sorting the queue while keeping contributor trust intact.

What to know before you use it

SlopGuard is helpful, but it is not an oracle. The site says its measured results were based on a small golden set of 25 cases: 13 real slop submissions and 12 legitimate contributions. At the default threshold of 50, it reported 100% precision and 92% recall, which means it missed at least one real slop case.

That matters because an AI slop PR/issue guard can still produce false negatives, and you should treat the score as decision support rather than final truth. The app also states that it uses heuristics only and no LLM key, so its behavior is not the same as a full model-based classifier.

A few practical caveats:

• It does not auto-close anything, so you still need to maintain a review workflow.
• It should not replace official maintainer judgment for borderline contributions.
• It is best used as a triage layer, not as a substitute for reading code, context, or contributor history.
• The published metrics come from a limited evaluation set, so real-world performance may differ.

If you need a GitHub App that helps you score, quarantine, and explain likely machine-made submissions while keeping the final decision human, SlopGuard is built for exactly that AI slop PR/issue guard workflow.