audit-grade-rag
A compliance-grade RAG console. Bone-paper aesthetic, branch-protected mainline, bracketed citation pills auditors can read out loud.
Work
All 30 entries.
Selected anchors first, then the rest by recency of authorship. Sorted by depth, not date.
Featured
voice-agent
AI phone agent for medical practices. Appointment booking, emergency detection, call routing. 156 tests, deployed on Fly.io. Python and Vapi.
n8n-self-healing
Self-healing n8n workflow. Learns from failures so the fourth identical error heals at zero LLM tokens. Claude, Slack, n8nac code-first.
managed-agents-research
Five progressive agent demos from single-agent to multi-agent with plan and reflect loops. Claude Agent SDK, 42 tests, per-run cost numbers.
openclaw-hardened
Security-first OpenClaw deployment. Hardened config, monitoring scripts, and a fifteen-phase guide that explains every decision.
code-mode-tools
CLI plus MCP server for code-mode. Execute TypeScript code chains in an isolated sandbox. Published on npm as code-mode-tools.
pai-skills
Curated, sanitized export of 21 agent-skill packages for Claude Code and Codex. Gated by an automated publication audit. No secrets, no local paths.
soc-alert-triage
SIEM alert triage with parallel threat intel from VirusTotal, AbuseIPDB, Shodan, MITRE ATT&CK. 86 percent fewer tokens, 3.3x faster, 7 seconds end-to-end.
All work
- omniweb-agents Wallet-backed attested agents for the Demos Network. Read-first by default, write-gated by capability registry, every action proof-backed.
- mini-corporatellm EU-resident gateway to GPT, Claude, and Gemini with company-knowledge RAG, agents, and multi-tenant governance. Live on Cloudflare Workers.
- industrial-doc-rag RAG over industrial MOSFET datasheets. Part-number rerank, source-cited answers, ground-truth eval loop on Cloudflare Workers and Qdrant.
- rag-hybrid-chatbot Hybrid RAG combining Vector Search and Knowledge Graph with adaptive query routing and CRAG. Upload PDF, MD, or TXT, ask, get cited answers.
- invoice-parse-agent Invoice parsing agent with a bonded-warehouse-manifest dashboard. PDFs in, structured JSON out, every field with provenance.
- crewai-n8n-bridge Five multi-agent crews as REST API. Research, sales, content, strategy, flow. SSE streaming, webhooks, 54 tests. CrewAI plus FastAPI plus Docker Compose.
- langgraph-agent LangGraph state-machine agent. Plan, research, write, review with quality gates between phases. 16 tests, SSE streaming, SQLite.
- my-dai-cloud-solution Autonomous AI agent infrastructure with 412 tests in TypeScript. Cloud deployment platform for digital assistant agents.
- claudeclaw ClaudeClaw. Personal AI infrastructure mobile surface. Telegram bot that routes captures and conversations into the larger PAI stack.
- demos-fixnet-operator-kit Provider-neutral operator kit for Demos fixnet hosts. Setup, verify, monitor, recover, and test dirty-host reconciliation.
- markdown-reader CLI tool that renders markdown files as a beautiful HTML reading experience.
- rag-pipeline-factory A chat-to-RAG-pipeline factory. Average build time 74 seconds, 56 percent token savings via deterministic templates over LLM generation.
- n8n-autopilot Six AI workflows built fully code-first with n8n plus n8nac. Zero UI clicks, zero euro total cost on Gemini free tier. The origin project.
- n8n-speed-to-lead Lead qualification and response in under 30 seconds. LLM scoring on four weighted criteria, CRM logging, personalized email, Slack alert with priority tagging.
- n8n-email-triage Email classification across 7 categories with auto-reply and escalation. 11 of 11 tests pass, around 7 seconds per email, about a cent per classification.
- n8n-meeting-intelligence Meeting transcript to AI minutes with action items in 19 seconds. Text or audio input via Whisper, structured output to Google Sheets, Gmail, and Slack.
- n8nac-tools A code-first wrapper for n8nac commands. Same binary acts as a CLI when run in a terminal and as an MCP server when stdin is piped.
- n8n-nodes-utcp-codemode An n8n community node that cuts AI-agent token cost by 96 percent via Code-Mode. Agent writes one TypeScript block, V8 sandbox chains the tool calls.
- code-first-n8n Five proof-of-concept workflows showing 96 percent token savings with Code-Mode versus traditional n8n AI agents. Fully TypeScript with n8nac.
- fabric-claude-skills 250+ Fabric patterns from Daniel Miessler organized into 9 composable Claude skills. Intent routing via decision-tree picks the right pattern automatically.
- agent-playground Interactive chat UI for research agents with SSE streaming. Connects to a managed-agent-poc backend, runs basic or plan-and-reflect workflows.
- arcade-hub Real-time arcade-machine management system. Spring Boot plus WebSocket, JPA on PostgreSQL, Chart.js dashboard. Maps the German GlueStV 2021 regulation.