AGENTIC ENGINE Research Reproducibility Intelligence
Audit Pipeline 0 events 0 evidence
score pending
Analysis Trace SSE + REST
No trace events yet.
Evidence live resolvers
No live evidence resolved yet.
Workspace dashboard

Agentic Reproducibility Engine

Submit papers, monitor multi-agent audits, inspect resolver-backed evidence, and package audit-ready reproducibility reports from one operational workspace.

Evidence-first audit orchestration

The engine turns a paper into validated evidence, reproducibility risk scores, consistency challenges, and an audit-ready report.

Every run streams structured reasoning artifacts, approved tool calls, resolver outputs, critique loops, and report provenance as the workflow executes.

Live Execution Stream SSE
Run an audit to stream model calls, tool calls, consistency checks, and report assembly here.
Recent audits

Library preview

Resolver Coverage
New audit

Submit a paper and configure autonomy

Paste an arXiv ID, DOI, URL, or paper notes. The frontend fills the live audit payload and the backend handles the agent/tool execution.

Audit results

Verifier-backed reproducibility report

Run an audit to replace this with live model, evidence, and report data.

Evidence Explorer
Audit library

Saved audit runs

Search completed local runs and saved drafts by title, arXiv ID, decision, or evidence source.

Benchmarks and insights

Evidence discipline for reproducibility scoring

Benchmarks expose resolver coverage, score distribution, tool availability, and AMD/Qwen deployment proof points for judges.

Score Distribution
Tool and deployment readiness