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 cover how teams coordinate work — from multi-agent dispatch and task routing to authorization models and incident protocols. Emphasis is on structural decisions that survive context loss, personnel change, and system scale.
Workflows
Multi-Agent Workflow
The Evaluator–Builder–Verifier coordination model and the Plan → Implement → Verify loop that prevents unchecked AI output from reaching production.
Plan, Implement, Verify
Why the sequence matters: how treating planning as a discrete, gate-controlled phase reduces rework and surfaces misaligned assumptions before implementation begins.
Mental Compilation
The practice of tracing code execution mentally before running it — and why it remains essential even when AI writes most of the code.
Debugging Distributed Systems
Observability-first debugging protocols for systems where failures cross service, network, and time boundaries.
Technical Debt Triage
A structured classification model for distinguishing tactical debt (acceptable) from structural debt (compounding) — and the criteria that determine intervention timing.
Identity-Based Task Dispatch
Why routing tasks to a role is insufficient. The architecture of workstation contexts, identity maps, and creation gates that enables auditable agent dispatch.
Process
Breaking Changes at Scale
A versioning and deprecation protocol that decouples producer and consumer release timelines without requiring synchronized deploys.
Multi-Layer Authorization
Defense-in-depth authorization: why perimeter checks are insufficient and how layered enforcement at the domain, service, and data tiers closes the gap.
ADRs as Architecture
Architecture Decision Records as executable governance artifacts — not just documentation, but the record that makes future changes auditable and reversible.
Performance Optimization Patterns
The measurement-first methodology: profiling before optimizing, and the patterns that consistently deliver the highest return per engineering hour.
CODEOWNERS as Agent Authority Boundary
How Git’s CODEOWNERS mechanism becomes a machine-enforceable governance layer controlling which AI agent may modify which service — structural authority, not instructional.
Kubernetes Operator for Multi-Tenant Provisioning
Six milestones from CRD schema to production: RAII-based secret injection via Rust Drop, annotation-driven blue/green upgrade coordination, per-tenant NATS JetStream stream provisioning, and air-gap distribution with digest manifests.
The Dogfooding Loop: Product Builds Product
How routing internal bug reports through the product being built — and wiring workflow automation so ticket transitions trigger task creation — closes the autonomous development loop without human intervention.
Dual-Write Migration: Agent Skills as the Control Plane
How routing agent skill writes to both legacy and new ITSM systems during migration eliminates the sync daemon entirely — and makes cutover a configuration change rather than a migration event.
Federated Skill Orchestration: Modular Agent Capabilities at Scale
How composing a control-plane core with selective domain modules (Rust, TypeScript, UX/CMS) prevents context bloat and rule contradictions as an agent organization’s skill surface expands.
Glorious Refinement: Post-Closure Quality Audit
A dedicated audit agent reviews closed tasks against six gates — correctness, structural integrity, DRY, resilience, performance/security, and evolutionary opportunity — one week after closure. The agent never modifies code.
Requirement Lifecycle: Working-Backwards Before Building
An eight-section Amazon working-backwards template plus a four-stage review chain (Skeptic → Architect → CISO → Approval) as a hard gate before any builder agent executes.
Founder Direction Protocol: Keeping CoS Informed When Founders Talk to Builders
A structured comment format that synchronizes direct founder-to-builder communication back to the Chief of Staff agent — preventing contradictory execution when the normal coordination chain is bypassed.
Infrastructure
Self-Hosted CI Pipeline
The operational model for running act_runner-based CI on self-hosted infrastructure — including the configuration patterns that prevent the most common runner failures.
Runbook and Failure Ledger
Why a failure ledger maintained alongside runbooks produces better incident response than runbooks alone — and the format that makes both useful under pressure.
Stateful Services on Kubernetes
The failure modes that cause silent data loss in PVC-backed workloads — and the operational protocol that prevents them at provisioning, runtime, and incident response.
Automated Lessons-Learned Capture
How a three-layer architecture — in-session log, session-end notification, nightly wiki write — solves the /tmp durability gap in AI agent workflows, combined with search-based parent task automation and dependency unblocking.
Three CI Security Gaps, One Sweep
Token-in-URL exposure, Cargo.lock exclusion defeating cargo audit on transitive dependencies, and shared agent credentials — three CI security gaps found and fixed in a coordinated same-day sweep across four Rust service repositories.