/work/12-factor-agentops

// DevOps for vibe coding

DevOps took a decade to figure out how to make infrastructure reliable, and SRE codified it. AI agents are the new indeterministic worker class. Making them reliable requires compiled context, validated output, and compounding knowledge.

12-Factor AgentOps adapts those hard-won patterns for AI delivery work. For measuring human-AI collaboration sessions, see vibe-check.

> Compiled context. Validated output. Compounding knowledge.

12

factors

3

tiers

2+

years battle-tested

// the 12 factors

FOUNDATION

get the inputs right

• I. Automated Tracking

• II. Context Loading

• III. Focused Agents

• IV. Continuous Validation

OPERATIONS

run it like a system

• V. Measure Everything

• VI. Resume Work

• VII. Smart Routing

• VIII. Human Validation

IMPROVEMENT

compound over time

• IX. Mine Patterns

• X. Small Iterations

• XI. Fail-Safe Checks

• XII. Package Patterns

// key insight

The translation table: every DevOps primitive has an AI-delivery equivalent.

DevOps primitive	AI delivery
Context windows	load balancing
Token budgets	resource quotas
Validation gates	health checks
Agent routing	service mesh

// the engine

KNOWLEDGE FLYWHEEL

Every session produces learnings. ao forge extracts them, ao inject surfaces them next session. Knowledge compounds instead of decaying.

BROWNIAN RATCHET

Embrace variance (parallel agents), filter aggressively (validation gates), lock progress permanently (merge). Chaos in, quality out.

The deeper thesis: context orchestration, not agent orchestration →

// read the methodology

The full methodology: all 12 factors with examples, failure patterns, and implementation guides.

12factoragentops.com →GitHub →