[prompt-analysis] Copilot PR Prompt Analysis - 2026-07-11

[github-actions[bot]]

Summary

Analysis Period: Last 30 days (ending 2026-07-11)
Total PRs: 1,000 | Merged: 783 (79.1%) | Closed: 207 (20.9%) | Open: 10

Prompt Categories and Success Rates

Category	Total	Merged	Success Rate
Refactoring	60	50	83%
Bug Fix	718	580	81%
Documentation	38	30	79%
Feature Addition	34	25	74%
Testing	124	88	71%
Other	16	10	62%

Prompt Analysis

✅ Successful Prompt Patterns

Common characteristics in merged PRs:

• Average prompt length: 183 words
• Most common keywords: workflow, actions, agent, fix, normalize, remove, spec
• WIP tag rate: 1.1% (very rare)
• File references in body: 72% include backtick-quoted file paths

Example successful prompts:

1. PR Add formal safe-update specification for compiler enforcement and manifest baselines #44997: Add formal safe-update specification for compiler enforcement and manifest baseline — specific scope, clear deliverable → Merged
2. PR Skip closed-PR follow-up workflows on Copilot branches #44967: Skip closed-PR follow-up workflows on Copilot branches — isolated problem, precise fix → Merged
3. PR Normalize progressive-disclosure guidance across reporting workflows #44960: Normalize progressive-disclosure guidance across reporting workflows — explicit consistency goal → Merged

❌ Unsuccessful Prompt Patterns

Common characteristics in closed PRs:

• Average prompt length: 182 words (nearly identical to merged — length is NOT a differentiator)
• Common indicators: [WIP] tags, batch fix, vague scope like "general improvement"
• WIP tag rate: 15.9% (~14× higher than merged)
• Placeholder bodies ("Thanks for asking me to work on this..."): present in 7.7% of closed PRs

Example unsuccessful prompts:

1. PR [WIP] Batch fix 7 detection misconfigured workflows #44938: [WIP] Batch fix 7 detection misconfigured workflows — WIP + batch + vague → Closed
2. PR Normalize report prompt style across daily/weekly workflows (h3+ headers + details guidance) #44943: Normalize report prompt style across daily/weekly workflows — broad normalization without specific files → Closed
3. PR Document engine.token-weights and engine.extensions in the concise frontmatter reference #44937: Document engine.token-weights and engine.extensions — documentation overlap with other work → Closed

Key Insights

• WIP tags are the strongest negative signal: PRs with [WIP] in the title are closed at a 14× higher rate than merged PRs. Copilot PRs should never be created as work-in-progress; complete the work before opening.
• Prompt length doesn't matter — scope clarity does: Merged and closed PRs have nearly identical average prompt lengths (183 vs 182 words). The differentiator is how clearly the prompt defines the problem boundary and expected outcome, not word count.
• Refactoring and bug-fix prompts outperform feature prompts: Focused improvement prompts (83% and 81% success) consistently outperform broad feature additions (74%). Prompts that reference a specific drift or regression perform best.

Recommendations

1. DO: Start the PR body with a concise problem statement describing the observed drift, error, or inconsistency — successful PRs read like bug reports, not feature requests.
2. DO: Reference specific files, frontmatter keys, or workflow names using backtick syntax — 72% of merged PRs include inline code references.
3. AVOID: Creating PRs with [WIP] tags or placeholder bodies — these have a 15.9% WIP rate among closed PRs vs 1.1% for merged, and placeholder bodies ("Thanks for asking...") appear 2× more often in closed PRs.

Historical Trends (Last 7 Days — by creation date)

Date	PRs Created	Merged	Success Rate
2026-07-05	66	57	86%
2026-07-06	49	38	78%
2026-07-07	43	28	65%
2026-07-08	54	43	80%
2026-07-09	45	33	75%
2026-07-10	41	32	78%
2026-07-11	40	21	68%

Trend: Today's in-progress rate (68%) is lower than the weekly average (~76%), which is expected as same-day PRs are still being reviewed. The overall 30-day rate (79.1%) is consistent with prior days (Jul 3–10 range: 79.1–80.5%), indicating stable Copilot PR quality.

30-Day Historical Summary (from cache)

Date	Total	Merged	Closed	Success Rate
2026-07-03	1000	801	178	81.8%
2026-07-04	1000	798	186	81.1%
2026-07-05	1000	805	191	80.8%
2026-07-06	1000	793	193	80.4%
2026-07-07	1000	792	198	80.0%
2026-07-08	1000	795	195	80.3%
2026-07-10	1000	795	200	79.9%
2026-07-11	1000	783	207	79.1%

Overall trend: slight decline in success rate (-2.7pp over 8 days). The rise in "closed" count (+29 PRs) warrants monitoring.

---

Generated by Copilot PR Prompt Analysis (Run: §29167857322)

References:

• §29167857322

Generated by 🔍 Copilot PR Prompt Pattern Analysis · 31.6 AIC · ⌖ 11 AIC · ⊞ 7.4K · ◷

• expires on Jul 12, 2026, 1:01 PM UTC-08:00