[prompt-analysis] Copilot PR Prompt Analysis - 2026-09-28

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2026-09-28

Summary

Analysis Period: Last 30 days
Total PRs: 571 | Merged: 444 (77.8%) | Closed: 122 (21.4%) | Open: 5 (0.9%)
Merge rate among completed PRs: 78.4%. No PRs had empty bodies. Categories are keyword-matched on titles, so "Other" is large.

Prompt Categories and Success Rates

Category	Total	Merged	Success Rate
Bug Fix	116	102	87.9%
Feature Addition	115	72	62.6%
Refactoring	14	12	85.7%
Documentation	19	12	63.2%
Testing	14	10	71.4%
Other (uncategorized)	293	236	80.5%

Prompt Analysis

✅ Successful Prompt Patterns

• Average body length: 190 words (median band 50–400 words)
• Prompts of 50–150 words: 84% merged; 150–400 words: 90% merged (of completed)
• Prompts with no file/path references (e.g. .go, .cjs, .md) merged 88% vs 64% with file references. This likely reflects task type, not causation (file-heavy PRs are larger).
• Action verbs: Fix, Track, Report, Surface, Ensure

Examples:

1. PR Track skill usage in audit artifacts #63688: Track skill usage in audit artifacts → Merged
2. PR Report MCP payload sizes in audit and aggregate analysis #63687: Report MCP payload sizes in audit and aggregate analysis → Merged
3. PR Fix slash-command PR prefetch caching and error handling #63678: Fix slash-command PR prefetch caching and error handling → Merged

❌ Unsuccessful Prompt Patterns

• Average body length: 338 words (vs 190 for merged)
• PRs with bodies over 400 words: only 35% merged (33 of 94 completed)
• Feature additions ("Allow...", "Add...") and docs changes close more often

Examples:

1. PR Improve AWF download failure diagnostics #63318: Improve AWF download failure diagnostics → Closed
2. PR Ensure a blank line follows closing </details> tags in generated markdown #63069: Ensure a blank line follows closing </details> tags → Closed
3. PR Allow add-labels max to set per-call label limit #63044: Allow add-labels max to set per-call label limit → Closed

Key Insights

• Length cliff: Prompts over 400 words merge at ~35% vs ~85–90% for 50–400 words. Long prompts likely signal sprawling scope.
• Bug fixes remain strongest (88%); feature additions are weakest (63%), likely due to ambiguous requirements.
• Merge rate has declined from 84.3% (2026-08-31 snapshot) to 78.4% today.

Recommendations

1. DO: Keep prompts between 50 and 400 words with one focused goal.
2. DO: Give features explicit acceptance criteria and the expected behaviour.
3. AVOID: Bundling multiple changes or very long context dumps (>400 words); split into separate PRs.

Historical Trends

Date	PRs	Success Rate	Top Category
2026-09-28	571	78.4%	Bug Fix (highest rate, 87.9%)
2026-08-31	994	84.3%	Bug Fix

Trend: Volume fell (994 → 571 PRs) and merge rate dropped ~6 points versus the last stored snapshot.

---

Generated by Copilot PR Prompt Analysis (Run: 36481232116)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · copilot · auto · 14.6 AIC · ⌖ 6.64 AIC · ⊞ 10.8K · ◷

• expires on Sep 29, 2026, 12:49 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Copilot PR Prompt Pattern Analysis.

A newer discussion is available at Discussion #64345.