ai-first-org-design-kit

AI-First Org Design Kit

Version: 1.5.1 | License: Apache-2.0 | Category: Organizational Design

Fourteen opinionated skills that guide founders and leaders through designing, deploying, adopting, and evolving organizations where agents handle coordination and execution while humans own specification and judgment.

This is not about adopting AI tools. It's about re-engineering the operating model.

The Thesis

Every organizational structure you work within is a fossil — a response to human constraints (scarce expertise, finite attention, expensive execution) that no longer exist. AI collapsed the cost of execution. The structures remain.

Organizations are specification machines that pretended to be execution machines. The value was always in defining what should exist and what "good" looks like. AI just proved it.

Who This Is For

• Founders building AI-first from scratch (greenfield)
• Leaders transitioning existing organizations (brownfield)
• CPOs/CTOs designing agentic infrastructure with governance
• Anyone asking "if agents do the execution, what do humans do?"

The Process

Skills run in sequence. Each feeds into the next.

Diagnose → Encode Identity → Specify → Design Gates → Architect Governance → Map Roles → Navigate Politics → Operationalize → Measure Adoption → Sprint → Deploy Agents → Evolve

Greenfield Path (Founders)

org-genome-builder → specification-writer → governance-architect → quality-gate-designer → role-value-mapper → operationalize → [optional] maturity-ladder → adoption-sprint-designer → agent-builder → [ongoing] evolution-auditor

Brownfield Path (Leaders)

coordination-audit → political-navigator → org-genome-builder → quality-gate-designer → specification-writer → role-value-mapper → governance-architect → operationalize → [optional] maturity-ladder → adoption-sprint-designer → usage-policy-writer → agent-builder → [ongoing] evolution-auditor

Already Deployed Path

holdout-evaluator → evolution-auditor → (revision skills based on findings) → operationalize → agent-builder

Driving Adoption Path

maturity-ladder → adoption-sprint-designer → usage-policy-writer → [ongoing] evolution-auditor

Quick Start

If building only two skills first: coordination-audit + org-genome-builder.

The Skills

All skills live under skills/ and follow the standard plugin SKILL.md format.

Skill	Your Specialist	What They Do
skills/coordination-audit	Organizational Diagnostician	Makes the invisible visible: where time actually goes, what's coordination vs. culture, highest-ROI encoding targets
skills/org-genome-builder	Org Psychologist + Systems Architect	Builds the organizational genome — values as decision rules, quality standards, communication norms, decision architecture
skills/specification-writer	Specification Engineer	Creates specs at any layer (task/workflow/governance/identity) precise enough for autonomous agent execution
skills/quality-gate-designer	Validation Architect	Converts approval chains into criteria-based quality gates with holdout-scenario validation
skills/governance-architect	Governance Systems Designer	Designs the full governance ecosystem — boundaries, escalation, policy generation, decision ledger, learning loops
skills/role-value-mapper	Team Architect	Designs roles from value flows and specification responsibility, not job titles
skills/political-navigator	Power Dynamics Strategist	Maps power structures, identifies resistance, creates reframes, sequences change
skills/operationalize	Operational Bridge	Distills all design artifacts into an agent-consumable primer (AGENT-PRIMER.md) and optionally merges governance into project CLAUDE.md
skills/holdout-evaluator	Quality Gate Judge	Validates agent work against hidden holdout scenarios using LLM-as-Judge, producing mapped feedback and gate telemetry
skills/evolution-auditor	Organizational Fitness Auditor	Runs the learning loop post-deployment: gate effectiveness, genome fitness, authority calibration, decision ledger
skills/agent-builder	Agent Configuration Engineer	Generates role-specific agent system prompts, tool permissions, and framework configs from organizational design artifacts
skills/maturity-ladder	Adoption Diagnostician	Builds per-role human AI maturity matrix using job titles or solo-founder operational modes, with barrier identification and barrier-informed progression paths
skills/adoption-sprint-designer	Sprint Architect	Designs structured adoption sprints with barrier-informed objectives, buddy pairing, flexible scheduling, demo format, and activity-based measurement
skills/usage-policy-writer	AI Policy Architect	Generates human-facing AI usage policy with approved tools, data classification, risk model reasoning, and exceptions

The router skill (skills/ai-first-kit) is the entry point — it diagnoses the user's situation and routes to the appropriate skill.

Core Concepts

All skills share a foundational vocabulary (see shared/concepts.md):

• Three-Variable Model — Organizational time is specification + coordination + execution (not planning vs. execution)
• Specification Stack — Four layers: Task → Workflow → Governance → Identity
• The Stranger Test — "Could someone with zero context produce acceptable output from this spec?"
• Dual-System Principle — Every structure serves coordination AND culture. Encode one, invest in the other.
• Five Resistance Archetypes — Gate Holder, Information Broker, Execution Expert, Empire Builder, Process Owner

Artifact Handoff

Skills save outputs to $HOME/.ai-first-kit/projects/{slug}/ (chmod 700) and automatically discover upstream artifacts. The slug is derived from the git repo root name for cross-project isolation. The coordination audit feeds into four other skills. The genome is referenced by everything downstream.

Installation

Via Synapti Marketplace (recommended)

git clone https://github.com/synaptiai/synapti-marketplace.git
claude plugin install ./synapti-marketplace/plugins/ai-first-org-design-kit

Project-level

cp -Rf path/to/synapti-marketplace/plugins/ai-first-org-design-kit .claude/plugins/ai-first-org-design-kit

Post-Deployment

Once operational, two additional skills maintain and extend the design:

• evolution-auditor — Monthly health check that runs the governance learning loop. Measures gate effectiveness, checks genome fitness against real decisions, calibrates the authority matrix, maintains the decision ledger, and routes findings to existing skills for revision.
• agent-builder — Generates role-specific agent configurations from your organizational design artifacts. Produces system prompts, tool permissions, and self-review checklists. Supports Claude Code, OpenAI Agents SDK, Anthropic Agent SDK, CrewAI, and custom frameworks.

Adoption

Three skills address the human side of AI adoption:

• maturity-ladder — Builds a per-role human AI maturity ladder with 4 levels (Not Engaged → Capable → Adoptive → Transformative) using human job titles or solo-founder operational modes — not agent definitions. Identifies the psychological barrier blocking each role and produces barrier-informed progression paths. Evidence-based assessment, not self-report.
• adoption-sprint-designer — Designs structured adoption sprints (2-3 day blocks or spread across a week) with barrier-informed objectives, buddy pairing, demo format, activity-based measurement, and leadership sequencing. When barrier data exists, selects the sprint objective that best addresses the dominant barrier. One sprint converts more than months of presentations.
• usage-policy-writer — Generates a human-facing AI usage policy explaining what tools are approved, data classification, and the reasoning behind each decision. Addresses the "I don't know what I'm allowed to do" barrier that kills adoption.

Claude Code Integration

The operationalize and agent-builder skills can generate Claude Code-native primitives in three layers:

Layer 1: CLAUDE.md @imports (Always-Loaded Foundation)

Select "CLAUDE.md with @imports" in operationalize to import genome foundation files directly:

@$HOME/.ai-first-kit/projects/{slug}/genome/00-identity/MISSION.md
@$HOME/.ai-first-kit/projects/{slug}/governance/HARD-BOUNDARIES.md
@$HOME/.ai-first-kit/projects/{slug}/genome/00-identity/VALUES.md

Content expands at session start and auto-updates when source files change. Always in context for every conversation.

Layer 2: Governance Skills (On-Demand Operations)

Select "Generate Claude Code governance skills" in operationalize to create five project-level skills:

Skill	Command	What It Does
org-record-decision	/org-record-decision	Append to the organizational decision ledger
org-novel-situation	/org-novel-situation	Draft a candidate policy for novel situations
org-voice-check	/org-voice-check	Review content against voice norms
org-gate-review	/org-gate-review [gate]	Self-review against a specific quality gate
org-values-check	/org-values-check	Check a decision against values and tradeoff rules

Skills read source files dynamically at invocation — they auto-update when upstream artifacts change without regeneration.

Review skills (org-voice-check, org-gate-review, org-values-check) run in forked context to avoid polluting the main conversation. Action skills (org-record-decision, org-novel-situation) run inline because they modify project files.

Layer 3: Sub-Agents (Role-Specific Agents)

Select "Register as Claude Code sub-agent" in agent-builder to create project-level agents:

# After registration, invoke via:
@"specification-architect (agent)" implement this feature
@"community-voice-guardian (agent)" review this article

Registered agents get:

• memory: project — Persistent learning across sessions (stored in .claude/agent-memory/)
• Preloaded governance skills via skills: field — governance operations available at startup
• Full system prompt inlined — self-contained, no file reads needed

Update Detection

Primitive	Auto-Updates?	Regeneration Trigger
CLAUDE.md @imports	Yes	Never (points to source files)
Governance skills	Mostly	Only on plugin version change (structural updates)
Sub-agents	No	Run agent-builder when roles, genome, or governance change

The evolution-auditor detects stale agents (upstream files newer than agent file) and recommends agent-builder re-runs.

What This Kit Does NOT Cover

• Tool selection. Which LLM, which agent framework. Tools are endpoints.
• Technical architecture. The kit bridges to agent consumption, not infrastructure.
• Legal/compliance. The governance skill helps design governance, not legal advice.
• Culture building. The kit identifies where culture needs design. What your culture should be is your job.

License

Apache-2.0. Build something.