Daily Prompt Clustering Analysis - Dec 30, 2025

[github-actions[bot]]

🔬 Copilot Agent Prompt Clustering Analysis

Analysis Date: 2025-12-30
Total PRs Analyzed: 986
Clusters Identified: 3
Overall Success Rate: 77.2%

Summary

This analysis uses NLP-based clustering to identify common patterns in copilot agent task prompts. We analyzed 986 pull requests and identified 3 distinct task clusters.

Most Common Task Types:

• Update: 428 tasks (43.4%)
• Add: 346 tasks (35.1%)
• Create: 224 tasks (22.7%)
• Modify: 218 tasks (22.1%)
• Test: 202 tasks (20.5%)

Highest Success Rate: Documentation Updates (79.4%)

Most Common Task Type: Feature Implementation (496 tasks, 50.3%)

Full Analysis Report

Cluster Analysis

Cluster 1: Feature Implementation

Size: 496 tasks (50.3% of total)

Success Rate: 77.2% (383/496 merged)

Top Actions: update, add, create

Keywords: update, add, version, cli, make, md

Characteristics:

• Average files changed: 16.8
• Average additions: 453 lines
• Average deletions: 297 lines
• Average comments per PR: 1.6
• Average reviews per PR: 1.6

Example PRs:

• Add minimal path format syntax reference to imports documentation #2097
• Add directory creation for copilot engine --add-dir paths #2099
• Add workflow status badges documentation page #2102

Sample Tasks:

1. Add minimal path format syntax reference to imports documentation
2. Add directory creation for copilot engine --add-dir paths

---

Cluster 2: Documentation Updates

Size: 282 tasks (28.6% of total)

Success Rate: 79.4% (224/282 merged)

Top Actions: update, add, create

Keywords: pkg, issue, files, githubnext, code, functions

Characteristics:

• Average files changed: 12.0
• Average additions: 477 lines
• Average deletions: 143 lines
• Average comments per PR: 1.3
• Average reviews per PR: 1.9

Example PRs:

• Refactor duplicate MCP code patterns for improved maintainability #2171
• [WIP] Comment on issue #2157 regarding recurrence failure #2209
• Fix: Raise error when agentic workflow hits max-turns limit #2235

Sample Tasks:

1. Refactor duplicate MCP code patterns for improved maintainability
2. Comment on issue regarding recurrence failure

---

Cluster 3: Feature Implementation

Size: 208 tasks (21.1% of total)

Success Rate: 74.0% (154/208 merged)

Top Actions: create, add, update

Keywords: workflow, agentic, workflows, github, add, file

Characteristics:

• Average files changed: 10.1
• Average additions: 1418 lines
• Average deletions: 216 lines
• Average comments per PR: 1.7
• Average reviews per PR: 2.1

Example PRs:

• Spread scheduled agentic workflows across 24 hours and add 6-hour schedules to smoke tests #2100
• Add smoke-outpost workflow for investigating failed smoke test runs #2103
• [WIP] Update variable name to 'log' and modify instructions #2107

Sample Tasks:

1. Spread scheduled agentic workflows across 24 hours and add 6-hour schedules to smoke tests
2. Add smoke-outpost workflow for investigating failed smoke test runs

---

Success Rate by Cluster

Cluster	Theme	Tasks	Success Rate	Avg Files	Top Actions
1	Feature Implementation	496	77.2%	16.8	update, add
2	Documentation Updates	282	79.4%	12.0	update, add
3	Feature Implementation	208	74.0%	10.1	create, add

Key Findings

1. Feature Implementation dominates: This cluster represents 50.3% of all tasks, indicating it's the most common type of work requested from copilot agents.

2. Documentation Updates has highest success: With a 79.4% merge rate, tasks in this category are most reliably completed successfully.

3. Feature Implementation tasks are most complex: Averaging 16.8 files changed, these tasks require the most extensive code modifications.

4. Workflow-related tasks: Cluster 3 shows a distinct focus on agentic workflow management, with high code additions (1418 lines average) suggesting these are substantial new features.

Recommendations

Based on the clustering analysis:

1. Optimize for Documentation Updates: These tasks show highest success rates. Consider creating specialized templates or guidelines for similar tasks.

2. Improve Feature Implementation: With only 74.0% success rate in Cluster 3, this task type needs better prompt engineering or agent capabilities.

3. Task Complexity Awareness: Tasks requiring more file changes have varying success rates. Consider breaking down complex tasks into smaller, focused subtasks.

4. Workflow Task Specialization: Given the prevalence of workflow-related tasks in Cluster 3, consider developing specialized agents or prompt templates for agentic workflow operations.

---

Methodology

Data Collection: Extracted task prompts from 986 copilot-created PRs in the last 30 days

Text Processing:

• Cleaned and normalized prompt text
• Removed code blocks, URLs, and special characters
• Applied TF-IDF vectorization with 100 features

Clustering:

• Used K-means clustering with optimal k=3 (determined via elbow method)
• Analyzed keyword patterns, action verbs, and success metrics

Analysis Metrics:

• Success rate (merge vs. close)
• Code complexity (files changed, additions/deletions)
• Collaboration metrics (comments, reviews)

AI generated by Copilot Agent Prompt Clustering Analysis