Generate SV-Bench metrics reports with summary.json and report.md.
npx skills add https://github.com/intertwine/security-verifiers --skill sv-report
Please help me install this Agent Skill. Skill: sv-report Source: https://github.com/intertwine/security-verifiers/tree/main/skills/sv-report Command: npx skills add https://github.com/intertwine/security-verifiers --skill sv-report
This Skill automates the generation and validation of performance metrics reports for evaluation runs within the SV-Bench framework, ensuring data integrity and comparability.
summary.json and report.md for individual evaluation runs.Generate a strict report for the E1 run located in outputs/evals/sv-env-network-logs--gpt-5-mini/your_run_id.
High-intent search queries and answers about installing and using this skill.
Generate SV-Bench metrics reports by processing evaluation outputs to produce summary.json and report.md files. The tool validates metrics contracts and creates comparison-friendly artifacts for E1 and E2 runs located in the outputs/evals/ directory.
Batch reporting processes multiple SV-Bench evaluation runs simultaneously. You can generate consolidated metrics reports across several runs at once, with options to filter the batch by specific environment or run ID.
Validate metrics contracts during report generation by enabling strict validation mode. This enforces adherence to defined metrics schemas for E1 and E2 evaluation runs, ensuring data integrity before producing summary.json and report.md artifacts.
Consolidate agent evaluation results into a human-readable report.md file for analysis and presentation. The tool transforms raw outputs from evals/ into comparison-friendly artifacts, making multiple network-logs environment runs easy to review.
No external dependencies are required to create SV-Bench summary.json files. The tool operates standalone using internal scripts to extract, validate, and format evaluation metrics from your existing outputs/evals/ directory structure.
SV-Bench report generation fails validation when evaluation outputs do not adhere to defined metrics contracts. Disable strict validation or ensure your E1 and E2 run data matches the required schema before generating reports.
intertwineExecute and analyze security verifier evaluations for network log anomaly detection and configuration verification.
rothwellmawela-cyberAnalyze web performance metrics and generate SEO audit reports.
lwyBZss8924dAutomate Harbor TB2 environment evaluations for ReCodeAgent workflows.
sourcegraphGenerate aggregate CodeScaleBench evaluation reports from Harbor runs in markdown, JSON, and CSV.