Retrieve the current evaluation session status, including evaluation count, detected frameworks, verdict history per file, and stability indicators, to understand what has been reviewed.
Preview which providers will be dispatched for a given expert and configuration, without making any provider calls. Use this to assemble a panel and then issue parallel requests for visible per-provider progress.
Full-featured MCP server for 1Panel server management panel. Provides 490+ tools for managing websites, databases, containers, files, and more through natural language.
Submit code for streaming evaluation that returns per-judge results progressively with running aggregate scores, providing early feedback as each of 45 judges completes.
Analyze only modified lines in a code diff by running all 45 judges on the full file and filtering findings to the specified line changes. Supports PR reviews and incremental analysis.
Execute the benchmark suite to generate a dashboard with breakdowns by judge, category, and difficulty. Identify weak spots and evaluate system quality using precision, recall, F1, and false positive rates.
Re-evaluate source code by incorporating developer context, disputed findings, and accepted rule IDs to adjust evaluation results with confidence filtering.
Run policy-aware code evaluation with domain-specific severity profiles, evidence calibration, and confidence scoring to ensure compliance with vertical requirements.