Real examples of AI agents in universities.https://agenticfeats.com/enhttps://agenticfeats.com/c/nsf-project-summary-drafting-cursor/https://agenticfeats.com/c/nsf-project-summary-drafting-cursor/The author used Cursor in Agent mode with Claude 3.7 Sonnet and two rules files to draft an NSF Project Summary from two papers. Result: The draft was usable but weaker than accepted NSF summaries; thinking mode helped, though it began hallucinating and misreading the sources. Evidence: Claim. Source: https://forum.cursor.com/t/anyone-use-cursor-for-academic-research-writing/57393Tue, 13 Oct 2026 15:29:03 GMTCursorResearchhttps://agenticfeats.com/c/synthetic-agi-attitudes-survey-study-manus/https://agenticfeats.com/c/synthetic-agi-attitudes-survey-study-manus/Manus planned a survey study on student attitudes to AGI, wrote a Google Apps Script for the form, simulated 1,000 students, and analyzed the data. Result: From one short prompt, Manus produced a publishable-style report with charts from 1,000 synthetic responses. The data is made up. Evidence: Shown. Source: https://futureofbeinghuman.com/p/when-agentic-ai-takes-charge-manusTue, 13 Oct 2026 15:29:03 GMTManusResearchhttps://agenticfeats.com/c/canvas-lms-tool-via-mcp-claude-code/https://agenticfeats.com/c/canvas-lms-tool-via-mcp-claude-code/Claude Code co-developed a Canvas LMS MCP server with an educator who is not a developer, covering courses, grading, analytics, and student data anonymization. Result: A roughly 3,000-line open-source Canvas tool built in four months. It cut a 45-minute weekly student check to a 2-minute conversation with Claude. Evidence: Shown. Source: https://chatwithgpt.substack.com/p/when-claude-code-becomes-your-codeveloperTue, 13 Oct 2026 15:29:03 GMTClaude CodeCodinghttps://agenticfeats.com/c/receipt-processing-in-cowork-claude-cowork/https://agenticfeats.com/c/receipt-processing-in-cowork-claude-cowork/Claude Cowork processed a PDF purchasing card receipt, extracting data and producing a cover sheet for the author's TPS report, running its own Python script. Result: Cowork ran the script, fixed an error itself, and returned finished output files with no terminal or manual dependency setup from the author. Evidence: Claim. Source: https://sites.duke.edu/ddmc/2026/02/24/what-claude-cowork-actually-does-and-what-it-doesntTue, 13 Oct 2026 15:29:03 GMTClaude CoworkFinance and Adminhttps://agenticfeats.com/c/course-modernization-with-agent-fleets-claude-code/https://agenticfeats.com/c/course-modernization-with-agent-fleets-claude-code/Claude Code turned course evaluations into tracked GitHub issues, then ran agents to rebuild assignments, exams and about two dozen slide decks. Result: Roughly two dozen decks were rebuilt into Quarto reveal.js with speaker notes and refresh checklists, and the approach was reused for two more courses. Evidence: Shown. Source: https://practicespace.substack.com/p/using-claude-code-to-modernize-myTue, 13 Oct 2026 15:29:03 GMTClaude CodeLearning