Copilot PR Prompt Analysis - 2025-12-22

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2025-12-22

Summary

Analysis Period: Last 30 days
Total PRs: 1000 | Merged: 745 (74.9%) | Closed: 250 (25.1%) | Open: 5

Prompt Categories and Success Rates

Category	Total	Merged	Success Rate
Bug Fix	979	741	75.7%
Feature	928	713	76.8%
Refactor	320	245	76.6%
Documentation	868	644	74.2%
Test	834	648	77.7%

Prompt Analysis

✅ Successful Prompt Patterns

Common characteristics in merged PRs:

• Average prompt length: 473 words
• Most common keywords: changes, workflow, github, agent, safe
• Top categories: bug fix, feature, refactor

Example successful prompts:

1. PR #7212: ## Fix Workflow Health Dashboard False Positive - Issue 🚨 P0: 48 Workflows Missing Compiled Lock Files #7105

• Analyze the reported 48 missing lock files
• Verify the compilation system ...

2. PR #7206: The release workflow generates documentation links in release highlights but couldn't access githubnext.github.io to verify paths.

Changes

• Add...

3. PR #7205: The issue reported that the Agent Performance Analyzer meta-orchestrator lacked GitHub MCP server configuration, preventing it from accessing GitHub A...

❌ Unsuccessful Prompt Patterns

Common characteristics in closed PRs:

• Average prompt length: 423 words
• Most common keywords: original, copilot, plan, changes, github
• Issues identified: Fewer detailed examples, different keyword patterns

Example unsuccessful prompts:

1. PR #7207: ## Plan: Fix agent-performance-analyzer to use local GitHub MCP

• Understand the issue - agent-performance-analyzer.md is missing mode: field i...

2. PR #7194: Dev Hawk was analyzing PR diffs and correlating errors with code changes. Now it inspects workflow run logs independently, focusing on error patterns ...
3. PR #7175: Evaluated whether map_helpers.go (2 functions) and git_helpers.go (1 function) should be consolidated, relocated, or kept as-is.

Decision: Kee...

Key Insights

• Testing prompts outperform average: Test-related prompts have a 77.7% success rate vs 74.9% overall, suggesting that prompts including testing considerations are more likely to be merged.
• Longer prompts correlate with success: Merged PRs have prompts averaging 473 words vs 423 words for closed PRs (11.7% longer), indicating more detailed prompts tend to succeed.
• Keywords unique to successful PRs: 'safe, workflow, agent' appear in top keywords for merged PRs but not closed PRs, suggesting these terms correlate with success.

Recommendations

Based on today's analysis:

1. Focus on test: This category has a 77.7% success rate with 834 PRs analyzed.
2. Provide comprehensive context: Successful prompts average 473 words. Include background, specific requirements, and success criteria.
3. Reference safe outputs: The keyword 'safe' appears frequently in merged PRs, suggesting prompts that reference safe output patterns are successful.

Historical Trends

Date	Completed PRs	Success Rate	Top Category
2025-12-22	995	74.9%	Test
2025-12-21	994	75.4%	Remove
2025-12-20	991	75.8%	N/A
2025-12-19	992	76.8%	Refactoring

Trend: Over the past week, success rates have ranged from 74.9% to 76.8%, with an average of 75.7%.

Full Keyword Analysis

Top 10 Keywords in Merged PRs

1. changes (331 occurrences)
2. workflow (310 occurrences)
3. github (209 occurrences)
4. agent (121 occurrences)
5. safe (117 occurrences)
6. test (113 occurrences)
7. mcp (108 occurrences)
8. workflows (108 occurrences)
9. cli (102 occurrences)
10. command (101 occurrences)

Top 10 Keywords in Closed PRs

1. original (80 occurrences)
2. copilot (71 occurrences)
3. plan (65 occurrences)
4. changes (65 occurrences)
5. github (59 occurrences)
6. make (54 occurrences)
7. workflow (53 occurrences)
8. description (53 occurrences)
9. issue (52 occurrences)
10. work (52 occurrences)

---

References:

• §20431195442

AI generated by Copilot PR Prompt Pattern Analysis

[github-actions[bot]]

This discussion was automatically closed because it was created by an agentic workflow more than 3 days ago.