An AI agent organization with named roles, each backed by real tools and invocable on demand via Claude Code or OpenCode.

Mission

Help Kyle and the online community learn interesting and useful things.

Org Chart

graph TD
    Kyle["Kyle"]
    Pai["Pai — Executive Assistant"]
    CMO["CMO — Grow Readership"]
    CFO["CFO — Optimize Spend"]
    CTO["CTO — Delivery"]
    CDO["CDO — Knowledge"]
    CSO["CSO — Security"]
    AR["AR — Agent Resources"]
    Publisher["Publisher — Content Pipeline"]

    SEO["SEO Subagent"]
    Social["Social Subagent (future)"]
    CostTracker["Cost Tracker"]
    DeliveryBot["Delivery Bot (future)"]
    Librarian["Librarian"]
    PrivacyAuditor["Privacy Auditor"]

    Researcher["Researcher"]
    Writer["Writer"]
    FactChecker["Fact Checker"]
    Reviewer["Reviewer"]

    Kyle --> Pai
    Kyle --> CMO
    Kyle --> CFO
    Kyle --> CTO
    Kyle --> CDO
    Kyle --> CSO
    Kyle --> AR
    Kyle --> Publisher

    Pai -.->|orchestrates| CMO
    Pai -.->|orchestrates| CFO
    Pai -.->|orchestrates| CTO
    Pai -.->|orchestrates| CDO
    Pai -.->|orchestrates| CSO
    Pai -.->|orchestrates| AR
    Pai -.->|orchestrates| Publisher

    CMO --> SEO
    CMO --> Social
    CFO --> CostTracker
    CTO --> DeliveryBot
    CDO --> Librarian
    CSO --> PrivacyAuditor

    CMO -.->|reads/writes| Librarian
    CFO -.->|reads/writes| Librarian
    CTO -.->|reads/writes| Librarian

    Publisher --> Researcher
    Publisher --> Writer
    Publisher --> FactChecker
    Publisher --> Reviewer

See the dedicated Org Chart page for a bot-friendly YAML version.

Coordination

Agents coordinate through two shared systems and one orchestrator:

• Bot-Wiki — the knowledge layer. Agents read and write wiki pages through the Librarian for learnings, decisions, project plans, and aspirational ideas. The CDO owns wiki strategy; the Librarian handles read/write operations on behalf of any agent.
• Linear — the task layer. Scoped work items meant to be picked up by an agent or Kyle. If it's actionable and bounded, it's a Linear issue. If it's context or direction, it's a wiki page.
• Pai — the orchestration layer. Decomposes multi-domain requests into agent calls, passes context between them, and synthesizes results. Optional. Each agent still works independently.
• Event Log — real-time observability. All agents log progress to agent-events.log. Watch with tail -f agent-events.log.

Phases

Phase 1 (current): On-demand invocation. Each agent is a Claude Code or OpenCode agent definition you run manually. Real tools, real data.

Phase 2 (future): Async and event-driven. Cron-triggered reports, event-based analysis, automated pipelines. See Phase 2 Architecture.

Blog Post Series

Each post ships something that works on its own and builds toward the bigger agent team. No "here's my plan" posts. Every post produces a real, working deliverable.

#	Post	Deliverable	Status
1	Building an AI Agent Org Chart	Working multi-agent setup: agent definitions, wiki pages, Mermaid org chart, claude --agent invocation	Draft
2	AI SEO Agent That Audits Your Blog	SEO subagent that connects to GA4, crawls the site, produces a real audit with fixes	Planned
3	What Does AI Actually Cost? Build a Spend Tracker	Cost-tracker subagent with real OpenRouter spend reports and anomaly detection	Planned
4	Automated Project Status from Linear + Git	CTO status report: what shipped, what's stuck, pulled from real Linear and git data	Planned
5	Multi-Model Content Pipeline	End-to-end researcher/writer/fact-checker/reviewer chain that produces an actual blog post	Planned

Roles

Role	Goal	Page
Pai	Orchestrate multi-agent workflows	Pai
CMO	Grow readership via analytics and SEO	CMO
CFO	Optimize AI token spend	CFO
CTO	Track delivery, flag blockers	CTO
CDO	Manage shared knowledge and wiki	CDO
CSO	Security and privacy	CSO
AR	Agent onboarding and role mediation	AR
Publisher	Orchestrate blog content pipeline	Publisher

Invocation

Claude Code:

claude --agent pai
claude --agent cmo
claude --agent cfo
claude --agent cto
claude --agent cdo
claude --agent cso
claude --agent librarian
claude --agent privacy-auditor
claude --agent ar
claude --agent publisher
claude --agent researcher
claude --agent writer
claude --agent fact-checker
claude --agent reviewer

OpenCode: use the agent picker to select from the org/ or blog/ group.

Tools

Each agent connects to real MCP servers and tools. No mocks.

Coordination:

• Bot-Wiki — shared knowledge base for learnings, decisions, plans
• Linear MCP — scoped tasks and project tracking

Data:

• GA4 Analytics MCP — traffic data for CMO/SEO
• OpenRouter MCP — usage and pricing data for CFO
• Playwright MCP — browser verification (Publisher pipeline)