[prompt-analysis] Copilot PR Prompt Analysis - 2026-10-04

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2026-10-04

Summary

Analysis Period: Last 30 days
Total PRs: 620 | Merged: 470 (75.8%) | Closed: 146 (23.5%) | Open: 4
Merge rate of completed PRs: 76.3%. No PRs had empty bodies.

Prompt Categories and Success Rates

Categorized by title keywords (first match wins).

Category	Total	Merged	Success Rate
Bug Fix	105	91	86.7%
Feature Addition	112	67	59.8%
Refactoring	15	12	80.0%
Documentation	24	18	75.0%
Testing	15	10	66.7%
Other/uncategorized	345	272	78.8%

Prompt Analysis

✅ Successful Prompt Patterns

• Average body length: 172 words (merged)
• Bodies reference specific files/extensions in 112/470 (24%) of merged PRs
• Action verbs: fix, guard, prevent, run, include

Examples:

1. PR Prevent Codex exec tool rejection with Copilot inference #65654: Prevent Codex exec tool rejection with Copilot inference → Merged
2. PR Guard Go allocation capacities against integer overflow #65653: Guard Go allocation capacities against integer overflow → Merged
3. PR Run persona evaluations through Claude Code sub-agents #65652: Run persona evaluations through Claude Code sub-agents → Merged

❌ Unsuccessful Prompt Patterns

• Average body length: 306 words (closed) — ~78% longer than merged
• File references in 76/146 (52%) of closed PRs (broad, multi-file scope)
• Features/new toolsets and cleanup-of-issues tasks close more often

Examples:

1. PR Close resolved Sergo lint issues #65618: Close resolved Sergo lint issues → Closed
2. PR Include safe-output failure details in logs exports #65616: Include safe-output failure details in logs exports → Closed
3. PR [mcp-tools] Add governance and automations toolsets; sync schema enum #65615: [mcp-tools] Add governance and automations toolsets; sync schema enum → Closed

Key Insights

• Narrow scope wins: bug fixes merge at 87% vs 60% for features.
• Longer prompts correlate with failure: closed PR bodies average 306 words vs 172 for merged, likely reflecting sprawling multi-file tasks.
• More file references ≠ success: closed PRs mention files twice as often (52% vs 24%), suggesting wide-touching changes are riskier.

Recommendations

1. DO: Scope prompts to one concrete defect with a clear expected outcome.
2. DO: Split large features into smaller, independently mergeable tasks.
3. AVOID: Long, multi-file prompts bundling schema, toolset and docs changes.

Historical Trends

Merge rate this period (76.3%) is lower than the earlier 2026-08-31 snapshot (84.3%, 994 PRs); volume is down, and the closed-PR body length grew (306 vs 144 words).

---

Generated by Copilot PR Prompt Analysis (Run: 37235332554)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · copilot · auto · 14 AIC · ⌖ 6.99 AIC · ⊞ 10.8K · ◷

• expires on Oct 5, 2026, 1:20 PM UTC-08:00

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-10-05T21:20:04.685Z.

Closed by Workflow