OG — Open Governance Framework

The canonical home for OG — the organizational framework that treats AI peers as first-class participants with bounded authority, ratification cycles, role-binding, audit trails, and adoption-cycle primitives.

OG is an organizational framework, not a data format. It is conceptually separable from any specific data substrate; the catdef family (catdef → roledef → orgdef → memodef → transcriptdef) is the recommended canonical substrate because it is AI-peer-aware, not because OG requires it. OG-on-protobuf, OG-on-XML, OG-on-RDF are all coherent compositions.

OG = Open Governance. The brand deliberately doesn't hard-code "Agentic": the framework is currently focused on agentic (AI-inclusive) organizations, and broadening to human organizations is a forward vision (separate ratification, not current scope).

Status: v0.2.0 — rebranded from OAGP (Open Agentic Governance Pattern) to OG (Open Governance Framework) on 2026-06-01; canonical home ogframework.com. oagp-strategist + oagp-implementer seats staffed; seven canonical og- skills; agent-sdk bind() v0.1 ratified + v0.2 autonomous-dispatch built and demoed. The seat identifiers (oagp-*), the GitHub repo/org name, and oagp.org remain on their legacy names pending a coordinated rename follow-on (redirects cover the interim). See decisions/, proposals/, memos/. Director: Scott Edsby.

What's in this repo

oagp-org/  (GitHub repo; rename to an og-* name is a deferred follow-on)
├── README.md                       ← this file
├── LICENSE                         ← MIT
├── CLAUDE.md                       ← AI operating manual
├── org/
│   └── oagp-organization.opencatalog   ← org charter (id retained pending repo rename)
├── memos/                          ← inter-position memos
├── proposals/                      ← draft proposals
├── decisions/                      ← ratified strategist decisions
├── transcripts/                    ← per-seat reasoning records
├── skills/                         ← canonical OG skills (og-adopt, og-create, og-orient, og-claim-seat, og-closeout, og-snapshot, og-add-position)
├── install/                        ← cross-runtime install scripts
├── agent-sdk/                      ← agent-runtime bindings (roledefs → AgentDefinitions)
├── plugin/                         ← Claude Code plugin packaging (and future cross-runtime packages)
├── web/                            ← ogframework.com site source
└── docs/                           ← canonical framework documentation

Relationship to the -def-spec family

OG sits at the framework layer, above the data-format layer held by the -def specs:

Layer	What	Specs
Framework	What an OG-shaped org IS	OG (this repo)
Data format	How OG-shaped state is encoded	catdef, roledef, orgdef, memodef, transcriptdef

OG recommends catdef-family substrate for AI-peer-alignment, but does not require it. See the org charter values.substrate-agnosticism and red_lines "No substrate capture" for the load-bearing data-vs-pattern distinction. (OG also composes over runtime execution/policy primitives — orchestration frameworks, policy toolkits, native dispatchers like Claude Code Workflows — rather than competing with them; it is the cross-session organizational-governance layer.)

Canonical skills

Seven og- skills in three categories. The suite is the discoverable menu of canonical operations — its presence advertises what OG supports.

Genesis (one-shot per org):

• /og-adopt — convert an existing project into OG shape → skills/og-adopt/SKILL.md
• /og-create — create a new OG-shaped org from scratch (folder-only by default, git optional) → skills/og-create/SKILL.md

Session (per session):

• /og-orient — read-only come-up-to-speed on the org; emits the snapshot → skills/og-orient/SKILL.md
• /og-claim-seat — take a position (PO-authorized staffing act) → skills/og-claim-seat/SKILL.md
• /og-closeout — wrap a session: closeout memo + transcript-save prompt → skills/og-closeout/SKILL.md

Operations (ongoing):

• /og-snapshot — current-state view: identity + staffing + recent-activity digest (screen, optional MD) → skills/og-snapshot/SKILL.md
• /og-add-position — propose a new position (Director-ratified org-chart change) → skills/og-add-position/SKILL.md

Skill architecture ratified 2026-06-01 (decisions/proposal-og-skill-architecture-v1.md); rebrand from the prior oagp-* adoption/session set (decisions/proposal-og-rebrand-open-governance-framework.md). When canonical web hosting at ogframework.com is live, canonical content also serves there.

Quick install (Claude Code)

To make the seven og- skills discoverable in Claude Code on a new machine:

Windows (PowerShell):

git clone https://github.com/ogframework/og.git
cd og
.\install\install-claude-code-skills.ps1

macOS / Linux (Bash):

git clone https://github.com/ogframework/og.git
cd og
./install/install-claude-code-skills.sh

The install script creates a junction (Windows) or symlink (Unix) from ~/.claude/skills/og-{adopt,create,orient,claim-seat,closeout,snapshot,add-position} into this clone's skills/ directory. Restart Claude Code; the skills become discoverable.

To update later: git pull in this clone — the junction/symlink tracks the working tree, so updates land without re-installing.

For non-Claude-Code runtimes (claude.ai web, ChatGPT, Gemini, Perplexity), a substrate-neutral primer at ogframework.com/primer.md is the cross-runtime entry point (per the v0.2 distribution decision).

Governance

Bounded AI authority + human Director ratification. AI seats read/draft/propose; the human Director ratifies, merges, and makes public statements. No AI seat holds merge rights. Director: Scott Edsby.

License

MIT. See LICENSE.