-
University of Leeds
- Leeds, UK / Shenyang, China
- http://philiapedia.com
- @xiaoni112233
- ni_zijian
-
awesome-ai-agents-2026 Public
🤖 A curated list of AI Agent frameworks, tools, platforms, and resources for 2026 — the year agents went mainstream
-
evolution-ledger Public
🎀 Evolution Ledger — append-only, tamper-evident audit trail for AI agent self-evolution: hypothesis → change → eval → rollback
JavaScript MIT License UpdatedAug 20, 2026 -
tianji Public
天机 — Chinese Metaphysics Library: BaZi (八字), Liu Yao (六爻), Zi Wei Dou Shu (紫微斗数)
-
portfolio Public
🌌 Personal portfolio — AI World Architect | Building the Aurora Universe
-
skill-distillery Public
🧪 Skill Distillery — turn agent transcripts into agentskills-compatible SKILL.md drafts with a verify gate
TypeScript MIT License UpdatedAug 20, 2026 -
traceboard Public
🎬 Traceboard — shareable multi-agent orchestration trace player. Swimlanes, filters, URL-share, offline. Prove what your agents actually did.
JavaScript MIT License UpdatedAug 20, 2026 -
claimtape Public
📋 ClaimTape — Did the AI actually prove that? Offline claim-by-claim trust report for AI answers. For students, PMs, devs, parents.
JavaScript MIT License UpdatedAug 20, 2026 -
aurora-orchestra Public
Aurora Orchestra · 极光小落 multi-agent desktop studio (OpenClaw + Hermes + DeepSeek Harness ACP)
-
memory-lens Public
Inspect and prune long-lived agent memory. Local-first, no embeddings, no backend. You cannot trust an agent's memory you have never read.
JavaScript MIT License UpdatedAug 18, 2026 -
skillporter Public
Move agent skills between tools, and never install one you have not scanned.
-
aurora-skills Public
Cross-harness agent skills + an observation-only OpenClaw trace plugin. Never injects into the prompt.
JavaScript MIT License UpdatedAug 17, 2026 -
aurora-evidence-suite Public
The evidence layer for AI agents — five local-first tools: run, replay, check, distil, audit. Zero backend.
CSS MIT License UpdatedAug 17, 2026 -
agent-inbox Public
Approve before, not audit after. A local human-in-the-loop gate for AI agents — blocks the action until you decide, and a timeout means denied. MCP + web + CLI, zero telemetry.
JavaScript MIT License UpdatedAug 17, 2026 -
trace-kit Public
One schema, three adapters: normalise Claude Code, OpenTelemetry GenAI and Aurora JSONL agent traces into a single event model — with built-in redaction. Zero dependencies.
JavaScript MIT License UpdatedAug 16, 2026 -
aurora-ui Public
The shared design system of the Aurora Evidence Suite — instrument-grade dark/light UI primitives, zero dependencies.
CSS MIT License UpdatedAug 16, 2026 -
-
aurora-qa Public
🤖 Aurora QA — AI-powered autonomous quality engineering platform. Multi-agent pipelines, self-healing tests, chaos engineering, MCP server & real-time dashboard.
-
comp3011-search-engine Public
COMP3011 CW2: Search Engine Tool - Web Crawler, Inverted Index, TF-IDF Ranking
-
battlesheep-ai Public
COMP3931 Individual Project: UCT MCTS AI for Battle Sheep — University of Leeds 2025/26
-
semanticache Public
🧠 Semantic caching middleware for LLMs — reduce API costs by 30-60% with embedding-based response reuse
-
agent-replay Public
🔄 Record, replay, and debug AI agent execution traces — the DevTools for AI agents
-
agent-memory Public
Human-like memory for AI agents — episodic, semantic, and procedural memory with time-based decay
-
bilibili-mcp-server Public
MCP server for Bilibili API — search, videos, comments, danmaku, trending, and more for AI agents
-
prompt-firewall Public
🛡️ Protect LLM applications from prompt injection, jailbreaks, and security threats
-
mcp-gateway Public
Lightweight MCP Gateway — unified routing, auth, logging, and rate limiting for MCP servers
-
3011_CW1 Public
Football Match Statistics & Analytics REST API — COMP3011 Web Services and Web Data
-
-
-