[prompt-analysis] Copilot PR Prompt Analysis - 2026-10-11

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2026-10-11

Summary

Analysis Period: Last 30 days (since 2026-09-11)
Total PRs: 573 | Merged: 485 (84.6%) | Closed: 85 (14.8%) | Open: 3

All PRs had a non-empty body.

Prompt Categories and Success Rates

Categorized by title keywords; PRs matching none are "Other" (361 PRs, 85.6% of completed merged).

Category	Total	Merged	Success Rate
Bug Fix	79	70	88.6%
Feature Addition	88 (2 open)	72	83.7% of completed
Refactoring	9	7	77.8%
Documentation	22	18	81.8%
Testing	11	9	81.8%
Other	364 (1 open)	309	85.6% of completed

Prompt Analysis

✅ Successful Prompt Patterns

• Average prompt length (merged): 157 words
• File references in body (.go/.cjs/.md/.yml/.js/.ts): 87 of 485 merged PRs (18%) – 85.3% success when referenced vs 84.9% when not (negligible)
• Common action verbs: fix, add, clarify, emit, update

Examples:

1. PR Clarify safe-output permissions and workflow trigger guidance #67737: Clarify safe-output permissions and workflow trigger guidance → Merged
2. PR Add race-detector coverage for concurrent Go packages #67623: Add race-detector coverage for concurrent Go packages → Merged
3. PR Fix routing smoke assertions: attribute agent traffic by model instead of nonexistent purpose values #67595: Fix routing smoke assertions: attribute agent traffic by model... → Merged

❌ Unsuccessful Prompt Patterns

• Average prompt length (closed): 182 words (about 16% longer)
• Titles in closed PRs often are open-ended "Investigate..." / "Prevent..." / "Restore..." tasks, or are still flagged [WIP]

Examples:

1. PR Investigate deterministic installs in compiled workflows #67469: Investigate deterministic installs in compiled workflows → Closed
2. PR [WIP] Restore classifier cost attribution in routed-run audits #67230: [WIP] Restore classifier cost attribution in routed-run audits → Closed
3. PR Prevent terminal safe-output loss on budget exhaustion #67470: Prevent terminal safe-output loss on budget exhaustion → Closed

Key Insights

• Bug fixes lead: 88.6% merged vs 77.8% for refactoring (small sample, n=9).
• Length: closed PR prompts are longer on average (182 vs 157 words), suggesting broad/multi-part scope hurts.
• Exploratory wording ("Investigate", WIP) correlates with closure; file references showed no measurable effect here.

Recommendations

1. DO: Scope each prompt to one concrete, verifiable change (a bug with an observable symptom).
2. DO: Keep prompts focused and concise (~150 words).
3. AVOID: Open-ended investigation or multi-concern tasks; split them into separate prompts.

Historical Trends

Date	PRs	Success Rate	Top Category
2026-10-11	573	85.1%	Feature (88)
2026-10-10	574	85.0%	Feature (85)
2026-10-09	575	84.7%	Bug Fix (90)
2026-10-08	552	84.3%	Bug Fix (90)

Trend: Merge rate is stable at ~84–85%; merged-prompt length has trended down (164 → 157 words).

---

Generated by Copilot PR Prompt Analysis (Run: 38173470578)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · copilot · auto · 13.7 AIC · ⌖ 0.688 AIC · ⊞ 10.9K · ◷

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

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-10-12T20:49:46.514Z.

Closed by Workflow