evidence-graph

the agent experience layer · observability + memory for AI coding agents (Claude Code + Codex) · local-first, typed, yours

Tauri 2 desktop intelligence platform — Cesium.js 3D globe, 185 live panels, MCP server for Claude Code, and an explainable algorithm intelligence layer (truth scoring, evidence graph, compound risk, forecast calibration) plus shortage / weather / insights engines. 600+ deterministic tests.

Agent skills for Gaia/LKM evidence-chain retrieval, evidence subgraph auditing, and scholarly review writing.

Turn messy artifacts into a cited knowledge graph and a queryable RAG packet, with baidu/Unlimited-OCR visual intake.

Evidence-first, auditable bibliometric research agent - from metadata to scientific maps and publication-ready reports.

VERA: The foundational verification and reliability sandbox for autonomous AI software engineering. Mathematically preventing LLM false-success.

A sourced web research agent on Cloudflare Workflows. Ask a question, follow evidence gaps, get a report with links.

Binary-Origami Figurator - evidence graph for investigative journalism with PROMPT scoring

Investigación asistida por IA sobre fuentes públicas, con grafos de conocimiento, procedencia verificable y revisión humana. Homenaje a Manuel Buendía.

Evidence-checked AI for deterministic private-credit verification.

Give your reverse-engineering agent a memory, a map, and receipts.

Local-first claim-evidence graphs and deterministic audit reports for AI-assisted research.

StepIn：面向零基础学生的实践优先 Learner Agent 平台，从真实简单任务开始，通过反馈、修订、迁移与证据链形成可验证能力。Practice-first learner agent for real-work simulation and capability evidence.

Local-first evidence maps for AI agents working across unfamiliar codebases.

Local-first evidence and analysis platform: typed source adapters, reproducible DuckDB products, evidence graphs, and agent-ready context.

AI research assistant for astronomy with reproducible data workflows, literature reasoning, evidence graphs, and provenance.

Open Employment Protocol — a reconciler, not a spec author. The candidate side and employer side each bring their own spec; OEP poses both arguments and settles a net outcome no single party owns. Built in the open, evidence-first. Protocol, not platform — the email of hiring.

Synthetic-data research platform for evidence-first longitudinal clinical trajectory analysis.

面向数学建模竞赛的证据驱动型 AI Agent：自动拆解问题、执行可复现实验、通过证据图谱追踪数值结论，并导出完整报告与独立复现项目包。

Local-first Neo4j research entity graph for indexing concepts, claims, evidence, and cross-project relationships.