[prompt-analysis] Copilot PR Prompt Analysis - 2026-10-17

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2026-10-17

Summary

Analysis Period: Last 30 days
Total PRs: 604 | Merged: 503 (83.3%) | Closed: 97 (16.1%) | Open: 4
Merge rate of completed PRs: 83.8%. No PRs had empty bodies.

Prompt Categories and Success Rates

(Categorized by title keywords; success = merged / (merged + closed).)

Category	Total	Merged	Success Rate
Bug Fix	108	94	87.9%
Feature Addition	91	73	80.2%
Refactoring	13	9	69.2%
Documentation	27	22	81.5%
Testing	20	17	85.0%
Other (bumps, workflow tweaks)	345	288	84.2%

Prompt Analysis

✅ Successful Prompt Patterns

• Average body length (merged): 161 words
• Frequent title terms: copilot, model, safe, agent, workflow, output, routing, daily, audit, queue
• Action verbs: fix, add, bump, require, support
• Bodies linking an issue (Fixes/Closes #) merged 84.5% vs 82.2% without.

Examples:

1. PR Add fallback pricing for the Copilot AOAI API-key smoke workflow #69746: Add fallback pricing for the Copilot AOAI API-key smoke workflow → Merged
2. PR Bump MCP Gateway to v0.4.35 #69581: Bump MCP Gateway to v0.4.35 → Merged
3. PR Bump gh-aw-firewall default to v0.28.62 #69526: Bump gh-aw-firewall default to v0.28.62 → Merged

❌ Unsuccessful Prompt Patterns

• Average body length (closed): 203 words (26% longer than merged)
• Frequent title terms: model, workflow, safe, engine, codex, prevent, route
• Issues: design/analysis or cross-repo-dependent work, opt-in/behavioral features, and refactors.

Examples:

1. PR Add opt-in duplicate PR cleanup to update and upgrade #69721: Add opt-in duplicate PR cleanup to update and upgrade → Closed
2. PR docs(specs): model routing analytics design #69507: docs(specs): model routing analytics design → Closed
3. PR Per-server HTTP MCP CA support requires gateway changes #69436: Per-server HTTP MCP CA support requires gateway changes → Closed
4. PR Sync gh-aw-actions pre-release status with gh-aw #69367: Sync gh-aw-actions pre-release status with gh-aw → Closed

Key Insights

• Length: Closed PR bodies are notably longer (203 vs 161 words); prompts under 50 words merged at 91.7% (n=12) vs 85.1% for >200 words (n=114, little difference) — the signal is weak but long bodies skew toward complex/ambiguous work.
• Category: Bug fixes are the most reliable (87.9%); refactors are weakest (69.2%, small n=13).
• File references in bodies made no meaningful difference (84.0% vs 83.8%); security/CVE titles (n=4) merged 75%.

Recommendations

1. DO: Scope prompts to a single, concrete bug or version bump with a linked issue.
2. DO: Break large design/feature work into smaller PRs rather than one long prompt.
3. AVOID: Tasks depending on changes in other repositories or open-ended design/analysis, and broad refactors.

Historical Trends

Date	PRs	Merge Rate
2026-10-17	604	83.8%
2026-08-31	994	84.3%

Trend: Merge rate is stable (~84%); volume is lower in this 30-day window (604 vs 994). Caveat: keyword categorization is heuristic and small categories are statistically noisy.

---

Generated by Copilot PR Prompt Analysis (Run: 39067741094)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · copilot · auto · 15.3 AIC · ⌖ 0.715 AIC · ⊞ 11.5K · ◷

• expires on Oct 18, 2026, 12:45 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 #70000.