Personal Command Infrastructure

Warren Command

A command layer for turning open-ended work into routed, concrete outputs. Specialized operators with clear boundaries. Research, synthesis, code execution, adversarial review, and scheduled follow-through — each in its own surface.

6
Production systems
95
Health score
24/7
Autonomous
Click to expand
coordination health queries events triggers + trust managed prompts memories context health data
Commands and queries from Warren via Telegram bot.
Telegram User Interface
14 repos, CI/CD, ops dashboards.
Web / GitHub Dashboards + CI
Ops notifications and Claude Code MCP tooling.
Slack / MCP Ops + Tooling
Control plane for 6 systems. Routes intelligence, coordinates services, surfaces health.
Cortix Autonomous OS Kernel :8100
Always-on Telegram agent. 9-mixin architecture, 3-tier model routing, hash-chained audit.
Kairos W Always-On AI Agent 4 systemd services
Real-time VPS monitoring. Service health, spend tracking, Telegram + Slack alerts.
Nerve Center Real-Time Intelligence :8091
Trust ledger, overnight pipeline, trigger engine. Governs all agent actions.
ANVIL Governance + Trust
Prompt lifecycle management. 9-step loop, 7 MCP tools, evidence-based promotion.
Prompt Foundry Prompt OS :8000
Semantic memory via Neon + pgvector. Cross-session context and identity persistence.
WCM Semantic Memory Neon + pgvector
DigitalOcean VPS Caddy (auto-TLS) systemd (12 units) Overnight Pipeline
Kernel
Agent
Governance
Prompts
Ops
Memory
Cortix port 8100
Unified Autonomous OS Kernel

The coordination layer that ties everything together. Routes intelligence queries, provides service state awareness, and maintains the health surface that every other system reports to.

FastAPI on port 8100
7 API endpoints
T1-T4 trust ladder
localhost-only, auth-gated
Kairos W 4 services
Always-On AI Agent

Telegram-controlled autonomous agent with 3-tier LLM routing. Decomposed from 4,337 lines to 473 via 9-mixin architecture. Hash-chained audit trail for every action.

9-mixin MRO pattern
3-tier model routing
bwrap sandboxed execution
Hermes 405B integration
ANVIL
Multi-Agent Governance

Trust ledger with scoring. Overnight pipeline evaluates triggers, generates briefings. Patterns graduate into reusable skills.

Prompt Foundry port 8000
Local-First Prompt OS

9-step operator loop from draft to evidence-backed promotion. 7 MCP tools for programmatic access. Prompts as first-class artifacts.

Nerve Center port 8091
Real-Time Intelligence

Live at ops.warrencommand.dev. VPS health monitoring, service status, spend tracking, and alert routing.

WCM Neon
Semantic Memory

Cross-session intelligence layer. Semantic search via pgvector. Identity persistence and context retrieval.

306
Build sessions
4,337
Lines decomposed to 473
12
systemd units
14 GitHub repos
242 tests
45% disk used
5 extracted modules
9 agent mixins
Most Polished

Prompt Foundry

Full operator loop from draft to evidence-backed promotion. 48 tests, Playwright E2E, 10 docs.

Live Dashboard

Nerve Center

Real-time VPS intelligence. Service health, spend tracking, alerting. Running in production.

Interactive Tour

Walkthrough

Visual walkthrough of the entire ecosystem. The best entry point for understanding how it all connects.

Stack

  • Python 3.11, asyncio
  • FastAPI
  • Telegram Bot API
  • React + TypeScript
  • Caddy reverse proxy
  • systemd services

Infrastructure

  • DigitalOcean (2 vCPU, 16GB)
  • Neon (Postgres + pgvector)
  • GitHub (14 repos)
  • Caddy (auto-TLS)
  • Cron overnight pipeline

Methodology

  • Session-based development
  • ANVIL governance
  • Hash-chained audit trails
  • Evidence-based promotion
  • Cross-model review (405B)
ESC or click to close