As a prompt engineer, I want code-analysis.md to be reviewed and refined so that it is maintainable and effective

[kiyotis]

Situation

code-analysis.md has grown to 685 lines — approximately 3× the size of comparable workflow files such as qa.md (243 lines). The file has been updated incrementally over time, with new instructions added in response to observed AI errors.

Pain

• Prompt engineers struggle to maintain the file: it is hard to see the overall structure, locate a rule, or assess the impact of a change
• The AI may not follow critical instructions reliably when the prompt is long, repetitive, or has conflicting statements
• It is unclear whether the current content reflects intentional design or accumulated patches

Benefit

• Prompt engineers can maintain code-analysis.md with confidence, knowing where rules live and that changes take effect
• The AI receives well-structured, unambiguous instructions and produces correct output consistently

Success Criteria

• code-analysis.md follows prompt engineering best practices (clear structure, no redundancy, authoritative single location per rule)
• The AI correctly follows the workflow instructions across representative test cases
• The file is easier to maintain: a prompt engineer can understand the full structure and locate any rule without reading the entire file