- Home
- Skills
- Research & Analysis
- paper integrity check
Works with the AI Agents you already use
paper integrity check
Scan academic manuscripts for AI signals, fabricated references, tortured phrases, and hidden text before submission.
$8
paper integrity check
Example session with this skill installed
Scan my paper 'neural_networks_v2.pdf' for AI signals and check if my references are real. Use the local papers in /previous_work for similarity.
- Read your context and instructions
- Compiled the paper integrity check
- Generated the document
I have completed the integrity scan of neural_networks_v2.pdf and compared it against the /previous_work corpus. I also verified the 12 citations against Crossref.
Integrity Report Summary
| Category | Finding | Severity |
|---|---|---|
| Hidden Text | No prompt injection strings or invisible characters found. | Low |
| References | 10 valid, 1 title mismatch (Ref #4), 1 DOI not found (Ref #9). | Medium |
| Tortured Phrases | Found "remains a mystery" and "extremely huge" in Section 2. | Low |
| Similarity | 8% lexical match with /previous_work/nn_v1.pdf. | Low |
| AI Signals | 3 paragraphs flagged (2 medium, 1 high). | Medium |
AI Signal Analysis
Paragraph 12 (Introduction): High signal (Score 4.0). ZipPy detected low burstiness; vocabulary density >4 per 100 words. Human Judgment: Reads as generic filler; recommend adding specific citations here.
Paragraph 24 (Methods): Medium signal (Score 2.5). Low sentence variance (0.28). Human Judgment: False positive likely; technical prose often follows this pattern.
Reference Discrepancies
Ref #4: DOI 10.1016/j.neunet.2023.01 resolves to "Deep Learning in Vision" but the paper lists it as "Neural Nets for Audio".
- Ref #9: DOI
10.5555/123456returned 404. Verify if this is a preprint or private draft.
Next steps
- Fix the title for Reference #4 and provide a working DOI for Reference #9.
- Rewrite Paragraph 12 in your own voice to reduce the AI-writing signal before submission.
- Replace the identified tortured phrases with standard technical terminology.
Note: This report indicates signals consistent with automated patterns; it is not proof of AI usage.
neural_networks_v2.pdf
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Academic integrity is increasingly difficult to verify with manual review alone. This skill provides a technical pre-submission scan for manuscripts, identifying signals of AI-generated text, citation errors, and hidden formatting issues that trigger journal rejection. It automates the tedious parts of integrity checking while maintaining a human-in-the-loop approach for final judgment.
What it does
- AI-Signal Detection analyzes paragraphs for low sentence-length variance, LLM-specific vocabulary density, and compression-based ZipPy signals.
- Reference Verification cross-checks DOIs and titles to identify fabricated, mismatched, or broken citations.
- Tortured Phrase Identification flags unusual word combinations often used to bypass plagiarism filters.
- Hidden Text Detection scans for prompt injection strings or invisible characters designed to manipulate automated reviewers.
- Similarity Spot-checking performs targeted web searches for exact-match strings to identify potential plagiarism.
How it works
- Prepare Environment by installing dependencies including ZipPy for lexical and burstiness analysis.
- Execute Scan by running the
check_paper.pyscript against the.pdf,.docx, or.texmanuscript. - Validate References by manually resolving DOIs the script could not verify and performing web spot-checks on flagged sentences.
- Render Report by re-processing the updated
report.jsonto generate a final HTML summary.
Frameworks & tools
This skill utilizes Python, LZMA compression analysis via ZipPy, Crossref API for reference validation, and optional GPU-based detection via Binoculars for advanced workflows.
Why this beats prompting it yourself
Standard prompts lack the specialized lexical analysis and reference cross-referencing capabilities required for academic rigor. This tool uses specific heuristics, like sentence-length coefficient of variation and N-gram matching, which are impossible to perform accurately through chat alone.
Use cases
- Pre-submission audit to ensure a thesis or journal article meets integrity standards.
- Reviewer screening to flag papers with high AI signals or fabricated citations.
- Misconduct investigation for identifying hidden text or prompt injections.
Known limitations
Requires Python and local dependency installation. Results are indicators, not proof of AI authorship. Network access is needed for reference verification.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.
Requires local Python environment and dependency installation. Cannot prove AI authorship. Web spot-checks require manual search steps. GPU required for Binoculars detection.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 13 days ago
- Accounts
GitHub
Crossref
Permissions
Allowed Hosts