The "Slop-Detection" Checklist for Maintainers

If a PR hits two or more of these flags, it is likely "AI Slop." You have full permission to close it without further review.

1. The "Polite Robot" PR Description

• The "Summarizer" Tone: Does the description simply restate the issue in perfectly grammatical, slightly repetitive prose? (e.g., "This Pull Request aims to resolve the reported issue by implementing the requested functionality in the designated module...")
• Generic Bullet Points: Are there bullet points that describe what the code does but not why the specific technical choices were made?
• Missing Context: No mention of local testing results, specific edge cases encountered, or "I tried X but Y happened."

2. Code Smells (The "Hallucination" Test)

• Non-Existent APIs: Does the code call functions or methods that don't exist in the library (e.g., calling circuit.apply_quantum_magic() in a Mitiq PR)?
• The "Stub" Fix: Does the PR "fix" the issue by adding a TODO, an empty try/except block, or by simply deleting the failing test case?
• Redundant Comments: Does the code have "Captain Obvious" comments?
  • Example: x = x + 1 # Increment x by one
• Inconsistent Logic: Does the code define a variable and then never use it, or import a massive library for a one-line change?

3. The "Quantum Reality Check"

• Physical Impossibility: Does the code attempt to perform operations that violate quantum mechanics or the specific constraints of the backend (e.g., mid-circuit measurement on a device that doesn't support it without checking)?
• Mathematical Nonsense: AI often struggles with complex indexing in tensors or state vectors. Check for shape mismatches that a human would have caught by running the code once.

4. Behavioral Red Flags

• The "Fastest Gun in the West": Was the PR submitted within minutes of the bounty being posted? (Unless it's a typo fix, it’s likely automated).
• Ghosting: If you ask a clarifying technical question ("Why did you choose this specific error mitigation technique over the default?") and they can't answer or provide another AI-generated paragraph, it's slop.

🛠 The "Fast-Close" Templates

Copy and paste these to save your team from "rejection guilt."

Template A: The "Likely AI" Rejection

"Hi @[Username], thank you for your interest in unitaryHACK and this issue. This PR appears to be a low-effort, AI-generated submission that hasn't been properly verified or tested against our codebase. Per unitaryHACK's Human-in-the-Loop AI Policy, we are closing this to prioritize maintainer time for substantive, human-led contributions. Please review the unitaryHACK Hacker Guide before submitting further PRs."

Template B: The "Explain Your Work" Challenge

(Use this if you aren't 100% sure but are suspicious)
"Hi @[Username], thanks for the PR. To ensure this meets our quality standards, could you please explain the logic behind [Specific Line/Function] and confirm that you've run the test suite locally? We require all hackers to be able to defend their technical choices to be eligible for bounties."

Pro-Tips for Maintainers:

• We recommend incorporating an LLM to oversee initial PR reviews for all reviews, several fellow maintainers have suggested CodeRabbit. This will cut down on you personally having to review submissions that may not be as serious.
• If you see a user submitting "slop" across multiple projects in the Unitary Foundation ecosystem, flag them in the #maintainers Discord channel. We can then issue a global ban for the duration of unitaryHACK to stop the noise.