Skip to main content

Documentation Index

Fetch the complete documentation index at: https://www.aidonow.com/llms.txt

Use this file to discover all available pages before exploring further.

Articles in this section document how Claude Code is used in a production autonomous development organization — not as a coding assistant, but as the execution layer for an AI-native engineering team. Patterns here are drawn from real deployments across multi-repo, multi-agent, multi-workstation environments.

CLAUDE.md & Memory

CLAUDE.md as Operational Constitution

How a single committed file becomes the persistent memory, governance layer, and behavioral contract for every Claude Code session — surviving context resets, agent handoffs, and personnel changes.

Memory Persistence Across Sessions

Three-layer memory architecture: in-session scratch, session-end wiki write, and nightly digest. How to design durable knowledge capture when the agent’s context window resets on every invocation.

Skills

Autoresearch: Nightly Intelligence with Local Inference

An 8-phase autonomous research loop running nightly at zero marginal cost — scoring GitHub Trending, HackerNews, arXiv, and competitor feeds with a local LLM, then auto-creating requirements from score-9+ findings.

Autoexplore: Zero-Cost QA Automation

How spawning a Playwright sub-agent to run 15 detection rules against a live dev environment — and triaging findings with Claude Code as the LLM at zero API cost — produces net-new bugs that manual review misses.

Hooks

PreToolUse Hooks as Architectural Guardrails

Hook-based enforcement removes memory from the equation by blocking non-compliant writes before they reach the filesystem. Financial type safety, multi-tenant query scoping, and boundary isolation — enforced at write time.

PostToolUse Hooks for Lifecycle Events

How PostToolUse hooks on Bash commands bridge Claude Code tool-use to external activity feeds — firing structured lifecycle events (task pickup, PR created, CI failed) without modifying any business logic.

MCP Servers

MCP Tool Routing: Nine Servers as an Agent Operating System

Routing 91 agent tools across nine domain-scoped MCP servers — with a capability-first pre-implementation gate — prevents reinvention and enforces organizational standards at the tool layer.

Whisper MCP Federation: Per-Actor Token Scoping

How scoping each AI agent’s activity-feed token to a single actor identity makes the audit trail tamper-evident — with graceful downgrade and a four-layer token storage model.

Agents & Subagents

Federated Skill Orchestration

Composing a control-plane core with selective domain modules prevents context bloat and rule contradictions as an agent organization’s skill surface expands.

Task Routing by Agent Identity

Why routing tasks to a role is insufficient. Workstation contexts, identity maps, and creation gates that enable auditable, deterministic agent dispatch.