the agent experience layer · observability + memory for AI coding agents (Claude Code + Codex) · local-first, typed, yours
Cross-layer evidence admission for ROS 2 latency: refuses to emit a root cause when evidence is incomplete, and blocks unreliable diagnoses from reaching config changes.
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 research RAG with a Laya decision engine. Discover papers, inspect source-linked claims and evidence graphs, and generate cited drafts with transparent model decisions.
Evidence-first, auditable bibliometric research agent - from metadata to scientific maps and publication-ready reports.
A sourced web research agent on Cloudflare Workflows. Ask a question, follow evidence gaps, get a report with links.
VERA: The foundational verification and reliability sandbox for autonomous AI software engineering. Mathematically preventing LLM false-success.
Dialectic Programming Language for the Industrial Workstations of the World (the pun, not the semantics): a unityped, homoiconic, reflexive language with no typechecker, where a Judgement Evidence Graph records what is believed about each term and why. Pre-alpha; Chapel bootstrap.
Give your reverse-engineering agent a memory, a map, and receipts.
Investigación asistida por IA sobre fuentes públicas, con grafos de conocimiento, procedencia verificable y revisión humana. Homenaje a Manuel Buendía.
Connect Automotive Cybersecurity Risk to Reality — automotive cybersecurity digital-thread & evidence-graph platform: TARA, requirements traceability, ISO/SAE 21434, UN R155/R156, SBOM, change impact analysis and cybersecurity release gates.
Evidence-backed application security analysis for a controlled FastAPI fixture: deterministic Python graphs, authorization-context candidates, traceable reports, and an interactive React/Cytoscape viewer. Runs on macOS, Windows, and Linux.
Git-native biomedical Research Graph for DeepSeek Harness, with literature-grounded expansion and auditable local project files.
Evidence-checked AI for deterministic private-credit verification.
Evidence-driven legal AI prototype with temporal provenance graphs, citation verification, review workflows, and offline evaluations.
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.
A provenance-first research knowledge graph for humans and AI agents. Local-first, SQLite-backed and standard library only. Free personal use under AL-1.1; commercial use under ACL-1.0.
Local-first claim-evidence graphs and deterministic audit reports for AI-assisted research.
To associate your repository with the evidence-graph topic, visit your repo's landing page and select "manage topics."