[prompt-analysis] Copilot PR Prompt Analysis - 2026-05-06

[github-actions[bot]]

Summary

Analysis Period: Last 30 days
Total PRs: 1000 | Merged: 780 (78.0%) | Closed: 214 (21.4%) | Open: 6 (0.6%)

Overall merge rate holds steady at 78%, consistent with the past two weeks. Bug fix prompts dominate at 64% of all PRs. Test-related prompts show the highest success rate this period.

Prompt Categories and Success Rates

Category	Total	Merged	Success Rate
Bug Fix	636	495	77.8%
Feature Addition	269	206	76.6%
Testing	35	31	88.6% ✅
Documentation	19	15	78.9%
Refactoring	16	11	68.8%
Other	25	22	88.0% ✅

Prompt Length Analysis

Outcome	Avg Word Count
Merged PRs	~1,267 words
Closed PRs	~1,248 words

Prompt length alone does not distinguish success — both merged and closed PRs have similar body lengths, suggesting content quality and specificity matter more than length.

✅ Successful Prompt Patterns (Merged PRs)

Common keywords in merged PR prompts:
github, command, actions, tool, http, block, tags, repos, object, triggering

Example merged PRs:

1. PR docs: move release process documentation to CONTRIBUTING.md #30682: docs: move release process documentation to CONTRIBUTING.md — "move release process documentation into CONTRIBUTING.md, remove the new RELEASES.md file" → Merged
2. PR docs: SPDD spec improvements — multiplier registry, safeguards, conflict norms, error norms, sync notes, compliance stubs #30681: docs: SPDD spec improvements — multiplier registry, safeguards, conflict norms — "Addresses all 10 P0/P1/P2 items from the daily SPDD work plan reviewing 5 specification files" → Merged
3. PR Stabilize template conditional fuzz assertions to eliminate malformed-input false failures #30678: Stabilize template conditional fuzz assertions — "The failing CI fuzz job was caused by overly strict assertions in FuzzWrapExpressionsInTemplateConditionals" → Merged

Characteristics:

• Clear references to specific files/modules
• Explicit problem statement with root cause
• Structured body (Summary, bullet points, checklist)
• Categorized commit prefix (docs:, fix:, etc.)

❌ Unsuccessful Prompt Patterns (Closed PRs)

Example closed PRs:

1. PR Add cache token multipliers and model cost information #30683: Add cache token multipliers and model cost information — Checklist-style body listing inspection tasks → Closed
2. PR Handle nested workflow sources and alternate source paths in Daily Workflow Updater #30635: Handle nested workflow sources and alternate source paths — "The Daily Workflow Updater only looked at top-level markdown files" → Closed
3. PR [WIP] Fix failures in agentic smoke test suite #30620: [WIP] Fix failures in agentic smoke test suite — "Thanks for asking me to work on this. I will get started on it and keep this PR's description up to date" → Closed

Red flags in closed PRs:

• [WIP] tag in title (signals incomplete work)
• Generic placeholder language ("I will get started on it")
• Pure checklist without narrative context
• Overly broad scope (multiple unrelated concerns)

Key Insights

• Test-related prompts lead in success rate (88.6%): PRs focused on adding or fixing tests are the most likely to be merged. Testing prompts tend to be specific and verifiable.
• Refactoring has the lowest success rate (68.8%): Broad refactoring PRs are harder to review and more likely to conflict or be superseded — consider breaking them into smaller, focused changes.
• [WIP] prefix strongly correlates with closure: Any PR marked as work-in-progress is significantly more likely to be closed. Avoid using [WIP] in Copilot-generated PR titles.

Recommendations

1. DO: Reference specific files, functions, or test cases in your prompt — concrete references lead to focused, reviewable PRs.
2. DO: Frame bug fix prompts with the root cause and failing behavior — context helps Copilot generate a targeted fix.
3. DO: Request test additions alongside fixes — test-focused prompts have the highest merge rate (88.6%).
4. AVOID: Using [WIP] or vague placeholder language — these signal incomplete work and lead to closure.
5. AVOID: Broad refactoring prompts — scope them to a single file or concern for better outcomes.

Historical Trends

Date	PRs	Success Rate	Top Category
2026-05-06	1000	78.0%	Bug Fix (78%)
2026-05-04	1000	78.0%	Bug Fix (79%)
2026-04-30	1000	77.8%	Bug Fix (78%)
2026-04-27	1000	78.0%	Bug Fix (78%)
2026-04-26	1000	78.1%	Bug Fix (78%)

Trend: Merge rates have been remarkably stable at ~78% for the past 10 days. Bug fix prompts consistently dominate volume. No significant shifts in category distribution.

---

References: §25460769779

Generated by Copilot PR Prompt Pattern Analysis · ● 3.3M · ◷

• expires on May 7, 2026, 9:04 PM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-05-07T21:04:16.370Z.

Closed by Workflow