Pricing
A review is billed when it runs: $5.00 in Single or Quick mode, $8.00 in Full mode (all 4 models). A failed review is refunded automatically.
Multi-AI Parallel Code Review
Master Code Review sends your code to up to 4 AI models in parallel (Claude Opus 5, GPT-5.6 Sol, Gemini 3 Pro, Claude Sonnet 5), then aggregates the results to surface consensus issues that multiple models agree on.
How It Works
- Upload a code file or paste code
- Select code type and focus areas
- Choose Full (4 AI), Quick (3 AI + Opus consensus), or Single (1 AI + Master-Mind approval)
- AI models review code in parallel
- View consensus findings and per-model results
AI Models
Primary: Claude Opus 5, GPT-5.6 Sol, Gemini 3 Pro
Secondary: Claude Sonnet 5
Review Categories
Security
Performance
Best Practices
Maintainability
ServiceNow
Quick Actions
Ctrl+V Paste code into the editor
Drop Drag files onto the upload zone
How a Review Runs — Flow of Events
- Provide code — upload one or more files, or paste code directly.
- (Optional) Attach context — expand the enrichment panels below the editor to add screenshots, a recorded session, ServiceNow record context, code dependencies, local files, or a live filesystem/ServiceNow connection (see Context Enrichment below).
- Choose options — Story number, Code Type, Focus Areas, and Execution Mode (Single / Quick / Full).
- Run Code Review — the selected models review in parallel. When a filesystem or ServiceNow connection is attached, models use live tools (MCP) to read related files and query the instance on-demand while they reason.
- Consensus & scoring — findings from all models are de-duplicated; issues agreed on by 2+ models become Consensus Issues. An overall 0–100 score is computed from severity.
- Review the report — Overall Score, Consensus Issues, Per-Model Results, Consolidated Report, and Collaboration Evidence. Export to PDF / DOCX / MD / TXT. Past runs are saved under History.
Execution Mode
- Single: Runs exactly one model that you pick from the AI Model dropdown (shown only in Single mode). Fastest / lowest cost.
- Quick: Runs the primary models for a faster pass — reports issues with 2+ model agreement.
- Full (Default): Runs all 4 models — the most thorough pass, reports issues with 2+ model agreement.
Context Enrichment Panels
Optional panels above Review Options let you give the models more context. Everything you attach is folded into the review.
- Screen Capture: Share a screen/window and capture screenshots (e.g. a form, error dialog, or diagram). Images are sent to the vision-capable models alongside the code.
- Session Replay: Record your interactions to give the models structured behavioral context.
- ServiceNow Record Context: Connect an instance and fetch records/portals/catalog items so their data accompanies the review.
- Code Dependencies: Paste a ServiceNow script to auto-detect and pull related Script Includes, tables, Business Rules, Client Scripts, UI Actions, and more.
- Local File Paths: Add absolute paths to local files/folders; their contents are read and included as context. Localhost only.
- Filesystem Access (MCP): Connect a local folder so models can read any file in it on-demand during the review — no file limit. Localhost only. Do not connect folders containing secrets or credentials.
Live Tools & MCP
When you connect a filesystem folder or a ServiceNow instance, the review models get live tool access (the same MCP capability as Master Mind) and call it autonomously while reviewing:
- Filesystem MCP: Models read related files on-demand (imports, Script Includes, callers) to verify findings against the real code instead of guessing. Available on localhost.
- ServiceNow MCP: Models run read-only queries against the connected instance to confirm artifact names, fields, ACLs, and sys_ids before reporting an issue. They can never modify records.
- If no connection is attached (or a tool call fails), the review automatically falls back to a standard text review.
Supported File Types
- Code: .js, .ts, .py, .java, .cs, .rb, .go, .rs, .php, .sql, .sh
- Web: .html, .css, .scss
- Config: .xml, .json, .yaml, .yml
- Documents: .docx, .xlsx, .pptx, .pdf (text extracted for review)
- Other: .txt, .md, .csv
Scoring
Scores are calculated from 0-100 based on issue severity. Consensus issues (found by 2+ models) carry higher weight. Severity levels: Critical, High, Medium, Low.