Current pairing
full/enhanced vs backbone/enhanced
Latest harvested pair stays reporting-comparable.
canonical public root
Truthful Copilot benchmark evidence, not fake parity.
This isolated presentation boundary keeps Copilot and Cursor in their own repo-native score models while making the shared view explicitly reporting-comparable.
The compare view keeps Copilot and Cursor in separate repo-native score models. Current shared rows are reporting-comparable observations, not mechanism-equivalent totals.
Capture skew
1.68 minutes
101 seconds between the latest Copilot and Cursor records.
Export shell
Next 16 static export
The proven Pages workflow and out/ build contract stay intact.
landing evidence rails
Visible routes and proof entry points stay above the fold.
Methodology and History remain first-class routes, while GitHub-hosted docs and benchmark notes stay one click away from the landing surface.
Route
Methodology
Inspect allowed comparability classes, repo-level provenance, and why reporting- comparable rows do not claim mechanism equivalence.
Open methodology →Route
History
Review benchmark runs, validation artifacts, and the generated manifest event in a single chronology.
Open history →Proof surface
Docs & repo proof
- README
Public repo overview, Copilot CLI-first scope, and current evidence framing.
- State contract
Boundary proof for repo-owned surfaces, install state, and validation expectations.
- References
GitHub-source-backed citations for Copilot host-product and comparison-scoped wording.
- Benchmark notes
Repo-native benchmark contract, harness notes, and release-gate context.
repo-native comparative readout
Sibling coherence without collapsing the scoring models.
Each card explains advantage inside its own harness first, then labels cross-host context as reporting-comparable rather than mechanism-equivalent parity.
oh-my-copilot
Repo-native baseline
Full enhanced records 100/100 in the repo-native scoring model.
reporting-comparable
9/9 required checks are passing with named proof links and timestamps. oh-my-copilot full/enhanced stays an observed repo-native benchmark row. It is safe for cross-host reporting, but it is not a mechanism-equivalent harness match.
oh-my-cursor
Repo-native baseline
Backbone baseline records 100/120 in the repo-native scoring model.
reporting-comparable
6/6 required checks are passing with named proof links and timestamps. oh-my-cursor backbone/baseline stays an observed repo-native benchmark row. It is safe for cross-host reporting, but it is not a mechanism-equivalent harness match.
state confidence
Named proof links remain visible alongside score summaries.
The flagship surface highlights required checks, timestamps, and proof links instead of turning benchmark output into marketing-only copy.
oh-my-copilot
State confidence for install, contract, and validation checks
oh-my-copilot starts by establishing a repo-native baseline before making any comparison claim. Cross-host reporting remains reporting-comparable and does not claim mechanism parity. 9/9 required checks are passing with named proof links and timestamps.
9/9 required checks passing
- Docs Validation — ok: research/omc-analysis.md has an Evidence section
- Power Validation — ok: VS Code settings enable AGENTS.md loading
- Root Validation — ok: post-tool hook logs root-workspace source
- Smoke Cli — GitHub Copilot CLI 1.0.34.
- Bootstrap — ok: cross-host methodology route explains comparability classes
- Install State — ok: installed source path is canonical: /home/zeyufu/Desktop/oh-my-copilot/packages/copilot-cli-plugin
- Standalone Hook Proof — ok: standalone workspace hook proof succeeded
- ROOT AGENT OK — GitHub Copilot CLI 1.0.34.
- PLUGIN AGENT OK — GitHub Copilot CLI 1.0.34.
oh-my-cursor
State confidence for auth, visibility, contract, and smoke checks
oh-my-cursor starts by establishing a repo-native baseline before making any comparison claim. Cross-host reporting remains reporting-comparable and does not claim mechanism parity. 6/6 required checks are passing with named proof links and timestamps.
6/6 required checks passing
- Default Auth — ✓ Logged in as fuzeyu09@gmail.com
- CURSOR MODEL AUTO OK — ✓ Logged in as fuzeyu09@gmail.com
- Surface Visibility — ok: docs/references.md
- State Contract — ok: .gitignore blocks speculative Cursor state files
- Backbone Verify — ok: AGENTS.md
- CURSOR AGENT OK — ok: reusing upstream default auth/model proof (environment-gated)
recent evidence
Generated history stays part of the public story.
These recent entries keep benchmark runs and validation artifacts visible from the homepage before a reader ever leaves the landing surface.
Apr 21, 2026, 3:38 PM · history
Cross-host harvest snapshot
3 observed rows were harvested with reporting-comparable semantics.
Proof: apps/cross-host-benchmark-site/generated/copilot-snapshots.json
Apr 21, 2026, 3:38 PM · history
Cross-host harvest snapshot
2 observed rows were harvested with reporting-comparable semantics.
Proof: apps/cross-host-benchmark-site/generated/cursor-snapshots.json
Apr 21, 2026, 4:53 AM · benchmark
Full enhanced benchmark
100/100 with 100/100 release gate
Proof: benchmark/results/current-full-enhanced
Apr 21, 2026, 4:53 AM · validation
Full enhanced benchmark report
100/100 with reporting-comparable comparability metadata.
Proof: benchmark/results/current-full-enhanced
Apr 21, 2026, 4:52 AM · benchmark
Quick enhanced benchmark
100/100 with 100/100 release gate
Proof: benchmark/results/current-quick-enhanced
Apr 21, 2026, 4:52 AM · benchmark
Backbone enhanced benchmark
120/120 with 120/120 release gate
Proof: benchmark/results/current-enhanced