[prompt-analysis] Copilot PR Prompt Analysis - 2026-10-30

[github-actions[bot]]

🤖 Copilot PR Prompt Pattern Analysis - 2026-10-30

Summary

Analysis Period: Last 30 days
Total PRs: 595 | Merged: 490 (82.4%) | Closed: 93 (15.6%) | Open: 12 (2.0%)

Merge rate among completed PRs: 84.1%. No PRs had empty bodies.

Prompt Categories and Success Rates

Categorized by title keywords (first match wins).

Category	Total	Merged	Success Rate (merged / completed)
Bug Fix	66	56	86%
Feature Addition	75	63	85%
Refactoring	10	7	70%
Documentation	28	25	89%
Testing	7	5	71%
Other	409	334	84%

Prompt Analysis

✅ Successful Prompt Patterns

• Average body length (merged): 161 words
• Documentation and focused bug-fix/feature prompts have the highest success (85-89%).
• Narrow, single-behavior titles, e.g. Include canonical issue IDs in duplicate close intents #73078 Include canonical issue IDs in duplicate close intents, Use topology-aware LLM gateway URL in detection and evals jobs #73046 Use topology-aware LLM gateway URL in detection and evals jobs, Route Claude Code User Documentation Review with model routing (cost / economy) #73032 Route Claude Code User Documentation Review with model routing → Merged

❌ Unsuccessful Prompt Patterns

• Average body length (closed): 135 words
• Many closed PRs are [WIP], "Investigate...", "Triage...", "Assess..." or "Define..." tasks (e.g. [WIP] Investigate recurring indexing and ignored-error findings #71881, [WIP] Fix failing GitHub Actions job 'agent' #69046, Investigate HydraFusion runtime model-policy rejection #68015, Assess Sighthound command-injection findings #67990, Triage unsubstantiated red-team findings #71621), which are exploratory with no concrete deliverable.
• Broad multi-area work: Refactor long workflow and CLI functions into focused helpers #71890 Refactor long workflow and CLI functions into focused helpers, Handle ignored errors and discarded results #71892 Handle ignored errors and discarded results, Reduce ambient context and enable targeted MCP log access #65568 Reduce ambient context....
• Large cross-cutting engine/model-routing or egress-policy changes (Fix pi alias model paths on arc-dind #72937, Exclude Copilot utility models from alias selection and restore detection fallbacks #72404, Narrow Chrome egress to browser-specific Google services #72101, Route model classification from workflow task text #64859/Route model classification from workflow task only #64841 duplicates) were often superseded or re-attempted.

Key Insights

• Pattern 1: Refactoring (70%) and testing (71%) prompts lag documentation (89%) and bug fixes (86%); the sample sizes are small (10 and 7), so treat these as weak signals.
• Pattern 2: Exploratory verbs (investigate, triage, assess) and [WIP] titles are over-represented among closed PRs.
• Pattern 3: Merged prompts are slightly longer (161 vs 135 words), suggesting more context and detail helps.
• Pattern 4: Duplicate/competing attempts (e.g. 64859 vs 64841) signal prompts that were re-run due to unclear scope.

Recommendations

1. DO: Specify a concrete deliverable and affected files/behavior, and include background context.
2. DO: Keep each prompt to a single, bounded change.
3. AVOID: Open-ended investigate/triage prompts and broad cross-cutting refactors without scoped acceptance criteria.

Historical Trends

Date	PRs	Merged	Closed	Open
2026-10-30	595	490	93	12
2026-10-29	612	503	98	11

Trend: Slight decline in volume with a stable merge ratio (about 82%).

---

Generated by Copilot PR Prompt Analysis (Run: 41124771935)

Generated by 🔍 Copilot PR Prompt Pattern Analysis · copilot · auto · ~15 AIC · ⌖ ~1 AIC · ⊞ 11.6K · ◷

• expires on Oct 31, 2026, 12:42 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 #73569.