[prompt-analysis] Copilot PR Prompt Analysis - 2026-07-12

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis — 2026-07-12

Summary

Analysis Period: Last 30 days
Total PRs: 1000 | Merged: 775 (78.9%) | Closed: 207 (21.1%) | Open: 18

Overall success rate (merged/completed): 78.9% — slightly below the recent 7-day average of ~79.4%.

---

Prompt Categories and Success Rates

Category	Total	Merged	Success Rate
Bug Fix	705	559	81.0%
Testing	178	132	75.4%
Refactor	17	13	76.5%
Documentation	42	30	71.4%
Feature Addition	41	30	73.2%
Other	17	11	64.7%

---

Prompt Analysis

✅ Successful Prompt Patterns

Common characteristics in merged PRs:

• Average prompt length: 192 words
• Most common keywords: workflow, actions, github, agent, copilot, engine, model
• Action verbs: fix, bump, add, resolve, correct

Example Successful Prompts

1. PR #45101: "fix: bump pinned gVisor release from 20250623.0 to 20250707.0" — Clear, specific version bump with root cause explained (release no longer exists on storage.googleapis.com). → Merged

2. PR #45025: "chore: bump firewall version to v0.27.30" — Specific version transition with inline TODO resolution noted. → Merged

3. PR #45006: "feat: Add docker-sbx runtime support to compiler" — Well-scoped feature addition with clear scope boundary (KVM-isolated Docker sbx microVM). → Merged

❌ Unsuccessful Prompt Patterns

Common characteristics in closed PRs:

• Average prompt length: 192 words (same as merged — length alone is not a predictor)
• Keywords associated with closure: wip, batch, draft, general, normalize

Example Unsuccessful Prompts

1. PR #44943: "Normalize report prompt style across daily/weekly workflows" — Broad scope (many files), vague change type ("normalization" across all workflows). → Closed

2. PR #44942: "Normalize dispatch-repository nested keys to kebab-case while preserving snake_case aliases" — Formatting-only change without a functional bug; no strong justification. → Closed

3. PR #44938: "[WIP] Batch fix 7 detection misconfigured workflows" — WIP label, large batch scope (7 files), no incremental path. → Closed

---

Key Insights

• Insight 1 — WIP signals failure: PRs with [WIP] or wip in the title/body are closed at ~15% rate vs ~1% for non-WIP PRs — a consistent multi-day pattern. Avoid WIP labels.
• Insight 2 — Bug fixes dominate and succeed: 70% of all PRs are bug fixes, with an 81% success rate — the highest single-category rate today. Specific, targeted bug reports lead to the best outcomes.
• Insight 3 — Broad "normalize/batch" prompts underperform: Prompts requesting sweeping style or formatting changes across many files (e.g., "normalize X across all workflows") are disproportionately closed. Scope-scoped, single-concern changes win.

---

Recommendations

1. DO: State the exact root cause or error in the prompt (e.g., "release 20250623.0 no longer exists") — this precision correlates with merged outcomes.
2. DO: Prefer bug-fix style prompts with a single, scoped change over broad refactoring requests.
3. AVOID: Using [WIP] or batch-scope language ("fix N issues", "normalize across all", "general improvement") — these consistently correlate with closed PRs.

---

Historical Trends (Last 7 Days)

Date	Total PRs	Success Rate	Top Category
2026-07-12	1000	78.9%	Bug Fix (81.0%)
2026-07-11	1000	79.1%	Bug Fix (81.0%)
2026-07-10	1000	79.9%	Bug Fix (80.3%)
2026-07-08	1000	79.5%	Bug Fix (81.2%)
2026-07-07	1000	79.2%	Refactor (83.3%)
2026-07-06	1000	79.3%	Bug Fix (81.1%)
2026-07-05	1000	80.5%	Bug Fix (81.6%)

Trend: Success rates are stable between 78.9–80.5%. Bug fix PRs consistently top success rates. No significant degradation this week.

---

Generated by Copilot PR Prompt Analysis (Run: §29208611323)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · 32.1 AIC · ⌖ 13.4 AIC · ⊞ 7.4K · ◷

• expires on Jul 13, 2026, 1:03 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 #45327.