Address review comments: remove fluff from AI guidelines

- Streamline ai-usage-guidelines.md to keep only verbatim key points
- Remove duplicate section headers
- Remove "Key requirements" section from AGENTS.md to prevent staleness

https://claude.ai/code/session_01CAnhr6MGdgQghH43KZYLpE

Author: claude

AGENTS.md

@@ -3,9 +3,3 @@
 AI agents contributing to this repository must read and follow the guidelines in the developer documentation:
 
 [Guidelines for AI-Assisted Development](doc/dev/ai-usage-guidelines.md)
-
-Key requirements:
-- A human manager must review and fully understand all contributed code
-- All commits must include the @username of the human manager
-- Pull requests require human manager self-review before peer review
-- Human manager effort must exceed effort requested from peer reviewers

doc/dev/ai-usage-guidelines.md

@@ -1,64 +1,23 @@
 # Guidelines for AI-Assisted Development
 
-This document provides guidelines for using AI tools (such as large language models, code generation assistants, and similar technologies) when contributing to the Empirical project.
-
 ## Guiding Principle
 
 **A human must ultimately take responsibility for all aspects of library code.**
 
-AI tools can be valuable aids in software development, but they do not replace human judgment, understanding, or accountability.
 All code contributed to Empirical, regardless of how it was generated, must meet the same standards of quality, correctness, and maintainability.
 
-## Guidelines
-
 ### 1. Human Manager Responsibility
 
-The human manager (i.e., the contributor submitting the code) is responsible for reviewing and fully understanding all contributed code.
-
-- You must be able to explain what the code does and why it works
-- You must verify that the code is correct, secure, and follows Empirical's coding standards
-- You must ensure the code integrates properly with the existing codebase
-- If you cannot fully understand or explain a piece of AI-generated code, do not submit it
+The human manager is responsible for reviewing and fully understanding all contributed code.
 
 ### 2. Commit Attribution
 
 Commits may be attributed to the AI tool used, but all commit messages should include the `@username` of the human manager.
 
-- This ensures traceability and accountability
-- The human manager remains responsible for the commit's contents
-- Example commit message format:
-  ```
-  Add utility function for data processing
-
-  Generated with assistance from [AI Tool Name]
-  Human manager: @username
-  ```
-
 ### 3. Pull Request Review Process
 
 Pull requests should go through the usual peer review process, with the human manager completing a pull request code review process first before requesting peer review.
 
-- Before adding the "Merge Ready" label, the human manager must conduct their own thorough review
-- Self-review should verify all items in the standard [code review checklist](contribution-guidelines-and-review.md#checklist)
-- Only after completing self-review should peer review be requested
-- Peer reviewers should be informed if AI tools were used in development
-
 ### 4. Effort Balance
 
 The effort contributed by the human manager should exceed the effort requested from peer reviewers.
-
-- The human manager bears the primary burden of ensuring code quality
-- Peer reviewers should not be expected to do extensive verification work that the human manager should have done
-- If a contribution requires significant review effort, the human manager should put in proportionally more effort to prepare and document the changes
-- This includes writing comprehensive tests, documentation, and explanatory comments
-
-## Guidelines
-
-When using AI tools to contribute to Empirical:
-
-1. **Understand** everything you submit
-2. **Attribute** AI involvement while maintaining human accountability
-3. **Review** your own work thoroughly before requesting peer review
-4. **Balance** effort so that you contribute more than you ask of others
-
-Following these guidelines ensures that AI tools enhance rather than undermine the quality and integrity of the Empirical codebase.