railguard-securityagentsonly-scan

To collect raw vulnerability findings, you can run a raw-agents scan that executes all 12 specialized security subagents unconditionally. This runs every agent in parallel without gates or triage enrichment to produce unfiltered results for benchmarking.

Unconditional security scanning serves as a benchmark control variant that produces raw findings. It enables direct comparison of coverage, recall, precision, and cost against gated orchestrated scans to evaluate scanning effectiveness across repositories.

No, this raw security scan explicitly disables gate-based discovery, data-flow tracing, false positive analysis, and triage enrichment. It focuses solely on complete subagent activation and raw findings collection for unfiltered benchmarking metadata.

You use unconditional agent scans to generate a consistent reference set of findings suitable for dashboarding and metric tracking. This approach provides a baseline to evaluate how gating and orchestration impact precision, recall, and cost.

The best way to benchmark precision and recall is to execute all 12 specialized security agents unconditionally against target repositories. This generates raw findings and benchmark-logging metadata, enabling direct comparisons between scan configurations.