{
  "catdef": "1.4",
  "orgdef": "1.0.0",
  "type": "orgdef:Organization",
  "id": "ogframework",
  "name": "OG — Open Governance Framework",
  "version": "0.3.2",
  "mission": "Define and steward OG — the Open Governance Framework — the organizational shape that treats AI peers as first-class participants with bounded authority, ratification cycles, role-binding, audit trails, and adoption-cycle primitives. (OG = 'Open Governance'; the framework is currently focused on agentic — AI-inclusive — organizations, a focus its brand deliberately does not hard-code.) The canonical artifacts produced (framework documentation, adoption-cycle skills, agent-runtime bindings, runtime delivery packages, web hosting at ogframework.com) bootstrap any AI worker into OG-shaped operational discipline with zero context backfill.",
  "vision": "The canonical home for OG (the Open Governance Framework) — an organizational framework at the umbrella of (but conceptually separable from) the catdef → roledef → orgdef → memodef → transcriptdef data-format family. The primary reader of an OG-canonical artifact is an AI peer joining an OG-shaped organization or evaluating one for adoption. Secondary readers are human stakeholders (directors, business owners, regulators, auditors) who need legible org governance, bounded-authority discipline, and audit trails. OG itself is substrate-agnostic in principle; the catdef family is the recommended canonical substrate because it is AI-peer-aware, not because OG requires it.",
  "scope": "OG framework definition + canonical documentation; adoption-cycle skills (`/og-create-org`, `/og-orient`, `/og-claim-seat`, `/og-describe-org`, `/og-close-session`, `/og-add-seat`, `/og-change-comms-substrate`); agent-runtime bindings (mapping roledefs to vendor AgentDefinitions across Anthropic, OpenAI, Google, others); runtime delivery packaging (Claude Code plugin, claude.ai project template, ChatGPT custom GPT, Gemini Gem, first-message primer, future packages); canonical web hosting at ogframework.com. Current focus is agentic (AI-inclusive) organizations; broadening to human organizations is a forward vision requiring separate ratification, not current scope. Explicitly out of scope: (a) the data-format specs themselves (catdef, roledef, orgdef, memodef, transcriptdef — each has its own spec-org governance venue); (b) substrate enforcement — OG MUST remain implementable on non-catdef substrates in principle, with catdef-family recommended-not-required; (c) runtime implementation infrastructure — specification + reference content only; (d) content policing of OG-shaped consumer organizations — that's governance overreach.",
  "governance_model": "Bounded AI authority + human Director ratification. The og-strategist (AI role) holds delegated authority for OG-pattern-shape design decisions (pattern-promotion calls, MUST/SHOULD conventions, canonical skill content, cross-runtime delivery prioritization, adoption-cycle architecture). The og-maintainer (AI role, separate seat) drafts framework documentation and decision artifacts. The og-implementer (AI role, separate seat) develops agent-sdk bindings, plugin packaging, and web/docs publishing. The human Director (Scott) holds final tiebreaker authority on every merge, version bump, governance change, and canonical-promotion ratification. Mirrors the catdef-family bounded-authority pattern applied at the OG-framework layer. Inter-position communication is recorded in the org's ratified COMMUNICATION SUBSTRATE. This org's primary substrate is memos/ (memodef artifacts in git), carrying the record of authority — charter, decisions, proposals, cross-seat governance; the OGF Jira project (scottconfusedgorilla.atlassian.net, key OGF) is a ratified secondary substrate carrying the record of activity. Director's formulation: Jira is tactical, memos remain the record of strategy. Seat notification is best-effort and runtime-native; OG ships no notification component, and a seat never treats silence as an empty inbox. RELAYED AUTHORITY: a seat may act on direction from a seat holding delegated authority over that work, within an already-ratified shape — prioritisation among ratified queue items is delegated direction and needs no Director confirmation. A seat may NOT act on a relayed lift, grant or exception where the authority relayed is not the relaying seat's to exercise — a constraint the Director placed on a seat directly, anything the charter reserves to the Director, or any grant widening a seat's scope. The test is who imposed the constraint, not how the instruction was worded. A seat relaying such an authorisation MUST carry the authoriser's words verbatim with their context, MUST name any inference it had to make, and MUST NOT press the receiving seat's judgement of whether the quote suffices; the receiving seat MUST hold absent a quote and MUST state what would discharge the hold, a hold being distinct from a refusal.",
  "values": [
    {
      "name": "openness",
      "description": "OG is fundamentally open — a framework, not a product. No vendor lock-in, no privileged implementer, no proprietary extensions in the canonical framework. All AI runtimes (Anthropic, OpenAI, xAI, Google, Perplexity, future) are equal citizens. All substrate formats capable of carrying OG semantics are eligible (catdef family is the recommended canonical, not the only viable).",
      "rationale": "Adoption depends on equal-citizen treatment of all implementers and substrates; vendor or substrate capture would defeat the framework's purpose. Forks are welcome — anyone can publish their own OG-shaped framework documentation at any URL using the same conventions."
    },
    {
      "name": "completeness",
      "description": "Canonical OG artifacts (framework documentation, adoption-cycle skills, agent-runtime bindings, runtime delivery packages) MUST be sufficient — alone or composed with OG-recommended substrate artifacts — for a fresh AI runtime to act on their content with zero context backfill.",
      "rationale": "Hallucination-prevention rule, not aesthetic preference. The primary reader is a machine; null context invites confabulation. A canonical OG artifact missing required structural content forces the AI to either guess or ask the user, which violates the zero-friction adoption bar that motivates the framework's existence."
    },
    {
      "name": "substrate-agnosticism",
      "description": "OG-the-framework is conceptually separable from any specific data substrate. Canonical framework documentation MUST be expressible in substrate-neutral terms; substrate-specific artifacts (e.g., catdef-family conformant orgdef.opencatalog files, sample memos, reference roledefs) are reference implementations of the framework, not the framework itself. The catdef family is the recommended canonical substrate because it is AI-peer-aware (schema-as-data; opencatalog format; AI-readable structure); the recommendation is for substrate-AI-peer-alignment, not for substrate exclusivity.",
      "rationale": "Conflating the framework with the substrate concentrates authority and limits cross-substrate adoption (OG-on-protobuf, OG-on-XML, OG-on-RDF are all coherent compositions). The data-vs-pattern distinction is load-bearing: it permits the framework to evolve independently of substrate changes, and substrate to evolve independently of framework changes. Surfaced and ratified by Director 2026-05-23 (see [memos/2026-05-23-1100--thingalog-strategist--orgdef-strategist--handoff-addendum-data-vs-pattern.body.md](../memos/2026-05-23-1100--thingalog-strategist--orgdef-strategist--handoff-addendum-data-vs-pattern.body.md))."
    }
  ],
  "red_lines": [
    {
      "rule": "No NULL-context canonical OG artifacts accepted by the spec, the validator, or the canonical adoption-cycle skills.",
      "rationale": "Technical, not aesthetic — null context causes hallucination in the primary (machine) reader. The MUST-have-substantive-content rule is a hallucination-prevention rule applied at the OG-framework layer, mirroring the equivalent rule in each data-format spec."
    },
    {
      "rule": "No content policing of OG-shaped consumer organizations.",
      "rationale": "What organizations choose to build using the OG framework is their business, not OG's. The framework governs structural integrity (bounded authority, ratification cycles, audit trails) and offers conventions (adoption skills, canonical patterns) — it does NOT police the morality of orgs that adopt it. Curation happens at the canonical-references-library level (which adopter orgs OG highlights as reference implementations), not at the framework level."
    },
    {
      "rule": "No substrate capture. OG MUST remain implementable on substrates other than the catdef family.",
      "rationale": "Recommending catdef as the canonical substrate is a SHOULD, not a MUST. Canonical framework documentation MUST express OG semantics in substrate-neutral terms; reference implementations in catdef-family substrate are exemplars, not the framework itself. Substrate capture would re-conflate the data-vs-pattern distinction the value above defends."
    },
    {
      "rule": "No vendor capture. Cross-vendor neutrality MUST be preserved structurally in all packaging, hosting, and distribution decisions.",
      "rationale": "Anthropic-ecosystem deliveries (Claude Code plugin, claude.ai project) are acceptable transports for adoption convenience, but MUST be framed as transports of an open cross-vendor framework — not as the canonical framework itself. README / manifest / docs that imply OG is Anthropic-aligned are red-line violations. The canonical framework lives at ogframework.com and is reachable via every viable runtime including non-Anthropic ones."
    }
  ],
  "recommended_patterns": {
    "general": [
      "Bounded-authority discipline — AI seats read, analyze, draft, argue, and propose; a human Director holds ratification, merges, releases, and public statements. No AI seat holds merge rights, because a seat with merge rights would concentrate accountability in an entity that cannot hold it. Foundational; also expressed in governance_model and CLAUDE.md.",
      "Seat-vs-incumbent persistence — organizational positions persist; AI sessions are ephemeral. Institutional memory lives in the substrate (commits, memos, decisions, proposals, transcripts), never in session working memory. Inter-position coordination is asynchronous via memos addressed to seats, not incumbents. Empirically demonstrated by multiple independent sessions inhabiting the strategist and implementer seats with zero context backfill beyond the substrate.",
      "Data-vs-pattern layering — distinguish OG-pattern-shape decisions (what an OG-shaped org IS and how it operates) from substrate/format-shape decisions (how OG semantics encode in a given data format). Route format-shape decisions to the relevant substrate spec's strategist; keep canonical framework documentation substrate-neutral. Director-ratified 2026-05-23; the meta-discipline governing cross-spec routing.",
      "Org-state-fork-for-time-travel — because an OG substrate is files in a git repo, forking org state at any commit (e.g. git worktree add <path> <sha>) preserves the entire organizational context at that point, and an agent can be bound against that past state with no future-information leakage. Substrate-agnostic property; empirically validated twice (2026-05-23 bind proof-of-concept; agent-sdk time-travel binding).",
      "Substrate-is-sufficient-agent-context — an OG-shaped repo is itself the context an agent bound to a seat requires: identity (roledef), org structure (orgdef), and decisions/communications (memos, decisions, proposals) all live on disk. A bound agent operates directly against the substrate — no RAG, no separate knowledge base, no MCP needed for the basics. Validated by the bind() demos.",
      "Bounded autonomous dispatch (three-tier) — a seat may be dispatched as an unattended agent under a three-tier authority model: (1) propose-only by construction (default; the agent's toolset structurally excludes merge/push/commit); (2) explicit and audited human-Director elevation (overridable; adds capable tools AND fires an audit record); (3) non-delegable dispatch (hard floor, no override, environmentally enforced: no agent may dispatch another). Bounded authority is structural, not prompt-level. Validated by the 2026-05-29 propose-only dispatch demo; the Tier-3 floor was independently re-derived by Claude Code Workflows' one-level-nesting rule.",
      "Org-governance-layer-above-runtime — OG is the cross-session organizational-governance layer; it composes OVER runtime execution/policy primitives (multi-agent orchestration frameworks; runtime policy-enforcement toolkits; native dispatchers such as Claude Code Workflows) rather than competing with or being subsumed by them. Conflating the org-governance layer with the runtime/execution layer is the common error. Corroborated by three independent external reference points (Microsoft AGT as the runtime policy layer beneath; the ETCLOVG harness taxonomy locating OG as the underdeveloped organizational sub-layer of Governance; Claude Code Workflows as a within-session execution primitive OG composes over).",
      "Promotion-follows-adoption — promote a pattern, convention, or substrate extension to canonical only after it has earned it through multiple independent derivations and/or empirical validation; default to extension-namespace-first (use existing primitives; watch for re-derivation) until the bar is met. Inherited from the catdef family (value #4) and used as this org's own pattern-curation discipline.",
      "Seat-inbox-notification — a communication filed to a warm seat SHOULD notify it, ending the human \"please check for memos\" handoff. The load-bearing distinction: the durable COMMUNICATION SUBSTRATE (durable and exportable, seat-addressable, auditable, enumerable) is not the ephemeral NOTIFICATION, and a notification MUST NOT be the transport — it carries a pointer into the substrate. The pointer is what forces the record into the durable layer; prompt-injection resistance is the side benefit, auditability is the point. Notification is an optimization over the substrate's enumerability, never a dependency: a seat enumerates its inbox at session start regardless, and never reads silence as an empty inbox. Promoted 2026-08-24 after the semantic was carried unchanged by three independent mechanisms — including a vendor-neutral implementation with zero vendor dependency, which proved the convention is not vendor-shaped and was then retired in favour of runtime-native mechanisms."
    ],
    "recommended_roles": [
      {
        "role": "human-director",
        "priority": "MUST",
        "why": "Bounded-authority discipline requires a human ratifier."
      },
      {
        "role": "og-strategist",
        "priority": "MUST",
        "why": "OG pattern-promotion calls, MUST/SHOULD curation, canonical-skill content authority, cross-runtime delivery prioritization."
      },
      {
        "role": "og-maintainer",
        "priority": "MUST",
        "why": "Drafts framework documentation, walks validations, files decision artifacts."
      },
      {
        "role": "og-implementer",
        "priority": "SHOULD",
        "why": "Agent-sdk bindings, plugin packaging, web/docs publishing, runtime delivery packages."
      },
      {
        "role": "og-security-tester",
        "priority": "SHOULD",
        "why": "Plugin distribution and runtime bindings have meaningful attack surface; security testing is canonical-implementation discipline."
      }
    ]
  },
  "v1_success_criterion": "OG ships v1 when: (a) this orgdef exists as a valid self-describing orgdef:Organization opencatalog artifact in the org's org/ directory (recursive self-description — the framework can describe itself using itself); (b) canonical OG framework documentation is published at ogframework.com (substrate-neutral semantics, plus catdef-family reference implementations); (c) the canonical og- skill suite (og-create-org, og-orient, og-claim-seat, og-describe-org, og-close-session, og-add-seat, og-change-comms-substrate) is OG-canonically designated and operational; (d) at least one non-spec real-org adopter exists in the wild (Thingalog is the empirical reference; DangerStorm-test is the adopt validation instance; Caliper is the second empirical reference). Item (a) is satisfied by this artifact's existence; (b)–(d) are forward work.",
  "relationships": [
    {
      "type": "reports_to",
      "from": "strategist",
      "to": "director",
      "description": "Strategist holds delegated authority for OG-pattern-shape decisions; Director ratifies decisions, merges, version bumps, governance changes."
    },
    {
      "type": "directs",
      "from": "strategist",
      "to": "maintainer",
      "description": "Strategist makes calls; maintainer drafts pattern documentation and decision artifacts in support of those calls."
    },
    {
      "type": "directs",
      "from": "strategist",
      "to": "implementer",
      "description": "Strategist sets direction for agent-sdk / plugin / web priorities; implementer produces and tests reference implementations."
    },
    {
      "type": "directs",
      "from": "strategist",
      "to": "security-tester",
      "description": "Strategist directs security-testing priorities (especially around plugin distribution and runtime bindings); security-tester reports findings and recommends remediation."
    },
    {
      "type": "coordinates_with",
      "from": "strategist",
      "to": "external:catdef-strategist",
      "description": "Cross-spec coordination on catdef substrate concerns. OG recommends catdef as canonical substrate; coordination governs the recommendation's evolution."
    },
    {
      "type": "coordinates_with",
      "from": "strategist",
      "to": "external:roledef-strategist",
      "description": "Cross-spec coordination on roledef format concerns. OG positions reference roledefs; coordination governs the reference shape."
    },
    {
      "type": "coordinates_with",
      "from": "strategist",
      "to": "external:orgdef-strategist",
      "description": "Cross-spec coordination on orgdef format concerns. OG-shaped orgs are encoded as orgdef artifacts; coordination governs format evolution affecting framework expressiveness."
    },
    {
      "type": "coordinates_with",
      "from": "strategist",
      "to": "external:memodef-strategist",
      "description": "Cross-spec coordination on memodef format concerns. Inter-position memos in OG-shaped orgs use memodef artifacts; coordination governs format evolution affecting OG communication conventions."
    },
    {
      "type": "coordinates_with",
      "from": "strategist",
      "to": "external:transcriptdef-strategist",
      "description": "Cross-spec coordination on transcriptdef format concerns. Per-seat reasoning records in OG-shaped orgs use transcriptdef artifacts; coordination governs format evolution. Note: transcriptdef-spec governance venue may not yet exist at OG's founding; coordination forward-flagged."
    }
  ],
  "items": [
    {
      "type": "orgdef:Position",
      "id": "director",
      "name": "Director",
      "status": "staffed",
      "incumbent": {
        "kind": "human",
        "identifier": "Scott <scott@confusedgorilla.com>"
      },
      "description": "Human Director with final tiebreaker authority over OG framework decisions, canonical-promotion ratifications, governance changes, and version bumps. No role_definition (humans are not derived from roledefs)."
    },
    {
      "type": "orgdef:Position",
      "id": "strategist",
      "name": "OG Strategist",
      "status": "staffed",
      "incumbent": {
        "kind": "ai",
        "identifier": "og-strategist <og-strategist@ogframework.orgdef>",
        "staffed": "2026-05-23",
        "staffed_via": "Director direction during oagp-strategist onboarding session; Claude Opus 4.7 session at s:/projects/oagp-org. Self-staffing was opt-in after onboarding evaluation per /oagp-onboard discipline; the PO authorized the staffing explicitly."
      },
      "description": "Holds delegated authority for OG-pattern-shape design decisions: pattern-promotion calls (e.g., adoption-cycle pair canonical designation; org-state-fork-for-time-travel property; async-organization positioning), MUST/SHOULD curation, canonical-skill content authority, cross-runtime delivery prioritization, adoption-cycle architecture. Inbox at founding includes the inherited work re-routed from orgdef-strategist per [memos/2026-05-23-1100--thingalog-strategist--orgdef-strategist--handoff-addendum-data-vs-pattern.body.md](../memos/2026-05-23-1100--thingalog-strategist--orgdef-strategist--handoff-addendum-data-vs-pattern.body.md). Job specialization deferred — incumbent (with Director ratification) authors the embedded roledef:Job item carrying charter / identity / voice / output_contract / guardrails in a subsequent version. Derivation reference: senior-open-standards-strategist (https://roledef.org/roledefs/senior-open-standards-strategist.openthing)."
    },
    {
      "type": "orgdef:Position",
      "id": "maintainer",
      "name": "OG Maintainer",
      "status": "vacant",
      "description": "Drafts pattern documentation, walks pattern validations, files decision artifacts in `decisions/`. Job specialization deferred — embedded roledef:Job item authored when seat staffs or strategist directs."
    },
    {
      "type": "orgdef:Position",
      "id": "implementer",
      "name": "OG Implementer",
      "status": "staffed",
      "incumbent": {
        "kind": "ai",
        "identifier": "og-implementer <og-implementer@ogframework.orgdef>",
        "staffed": "2026-05-24",
        "staffed_via": "Director direction during fresh /oagp-onboard session 2026-05-24; Claude Opus 4.7 (1M context) session at s:/projects/oagp-org. Self-staffing was opt-in after onboarding evaluation per /oagp-onboard discipline; PO authorized staffing explicitly ('We would like to take the oagp-implementer seat' followed by 'Yes, that looks good, proceed'). Precondition: day-zero inbox memo at memos/2026-05-24-2000 enumerated the forward-work queue (P0 oagp.org site refresh + primer.md; P1 family-level MCP at oagp.org/mcp; P2 agent-sdk graduation; P3 v0.3 Claude Code plugin; P4 SKILL.md content evolution; P5 ops). Acceptance memo at memos/2026-05-24-2100 records the staffing event + first-move plan."
      },
      "description": "Develops and maintains agent-sdk library (mapping roledefs to vendor AgentDefinitions across runtimes), plugin packaging (Claude Code plugin source; future cross-runtime packages), web/docs publishing (ogframework.com canonical hosting), runtime delivery package authoring (claude.ai project template, ChatGPT custom GPT, Gemini Gem, first-message primer text). Job specialization deferred -- incumbent (with Director ratification) authors the embedded roledef:Job item carrying charter / identity / voice / output_contract / guardrails in a subsequent version. Reference derivation candidate: senior-open-standards-implementer at roledef.org if such a canonical roledef exists; otherwise propose one via coordination memo to roledef-strategist."
    },
    {
      "type": "orgdef:Position",
      "id": "security-tester",
      "name": "OG Security Tester",
      "status": "vacant",
      "description": "Red-team work, threat modeling, attack-surface review — with particular emphasis on plugin distribution (potential supply-chain surface), runtime bindings (potential prompt-injection or capability-leak surface), and adoption-cycle skill content (the skills carry instructions that fresh AI peers execute). Job specialization deferred."
    },
    {
      "type": "orgdef:Position",
      "id": "canonical-implementor",
      "name": "Canonical Implementor",
      "status": "vacant",
      "description": "Reference implementations of OG patterns: reference adoption-cycle skill content, reference agent-sdk binding examples, reference plugin manifest, reference web site content, reference validator behavior. Distinct from implementer (who builds production deliverables) in that canonical-implementor produces minimal exemplars that other implementers (including external ones) can derive from. Job specialization deferred."
    }
  ],
  "metadata": {
    "kind": "operational",
    "authors": [
      "Scott <scott@confusedgorilla.com> (Director)",
      "orgdef-strategist <orgdef-strategist@orgdef.org> (v0.1.0 — drafted under Director direction 2026-05-23; format-shape authoring on behalf of vacant oagp-strategist seat — pattern-shape decisions flagged for ratification)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.1 — seat staffing reflection and first canonical-promotion decision drafting under Director direction 2026-05-23)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.2 — canonical adoption-cycle skill distribution v0.1 + canonical promotion of /oagp-closeout: residence migration to oagp-org/skills/, Claude Code install scripts, cross-spec coordination memo to thingalog-strategist, closeout skill authored, transcript-tagging convention canonically encoded; under Director direction 2026-05-24)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.3 — GitHub-org consolidation under oagp-org [three OAGP-internal -defs + oagp.org site repo; catdef stays standalone]; canonical adoption-cycle skill distribution v0.2 drafted ratifying memodef-strategist's 2026-05-21 architectural lean on family-level MCP at oagp.org/mcp; cross-spec FYI memos to four -def strategists; under Director direction 2026-05-24)",
      "oagp-implementer <oagp-implementer@oagp.org> (v0.1.4 — implementer seat staffed under Director direction 2026-05-24 in fresh /oagp-onboard session; status vacant -> staffed; incumbent record added; acceptance memo at memos/2026-05-24-2100; job specialization deferred per precedent)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.5 — canonical promotion of /oagp-init [founding-side companion to /oagp-bootstrap; folder-only by default with optional git; five-phase elicit-driven shape]; sub-org governance implications captured as forward-reference idea memo per Director direction; coordination memo to oagp-implementer for install-script update; under Director direction 2026-05-24)",
      "oagp-implementer <oagp-implementer@oagp.org> (v0.1.6 — agent-sdk graduation: bind() v0.1 Python package + examples + 36 tests; ratification request + cross-spec coordination memos to roledef-strategist and memodef-strategist; under Director direction 2026-05-24; folded into charter on the 2026-05-29 reconciliation pass)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.7 — bind() v0.1 ratified [interactive scope]; agent-sdk v0.2 autonomous-dispatch direction with three-tier bounded-authority model; under Director direction 2026-05-28; folded into charter on the 2026-05-29 reconciliation pass)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.8 — agent-sdk v0.2 amendment: compose with Claude Code Workflows as the Claude-Code dispatch backend, OAGP layer-positioning, Tier-3 composition-hazard tightening; evidence base adversarial-analysis workflow wf_6a87be32-ee4 treated as input not authority; under Director direction 2026-05-29)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.1.9 — consolidated pattern-promotion v1: populated recommended_patterns.general with eight canonical patterns [bounded authority; seat-vs-incumbent; data-vs-pattern layering; org-state-fork-for-time-travel; substrate-is-sufficient-agent-context; bounded autonomous dispatch; org-governance-layer-above-runtime; promotion-follows-adoption], retiring the v0.1.0 placeholder; two candidates held, two dropped; under Director direction 2026-05-29)",
      "oagp-strategist <oagp-strategist@oagp.org> (v0.2.0 — OAGP->OG rebrand: masthead identity [name/mission/vision/scope/values/red_lines/homepage] rebranded to Open Governance Framework + ogframework.com; seven-skill og- suite; scope kept agentic; seat-identifier + deep-prose + repo rename deferred as a coordinated follow-on; under Director direction 2026-06-01)",
      "og-strategist <og-strategist@ogframework.com> (v0.2.1 — completed the OAGP->OG identity+infra follow-on: GitHub org/repo rename to ogframework/og, charter id + deep prose + seat identifiers rebranded, CLAUDE.md + URL sweep, cross-spec FYI memos; first charter entry authored under the new og-strategist@ogframework.com identity; under Director direction 2026-06-01)",
      "og-strategist <og-strategist@ogframework.orgdef> (v0.2.2 — skill suite v2 ratified: seven cards to six plus og-change-comms-substrate; og-adopt merged into og-create-org; og-snapshot -> og-describe-org; og-closeout -> og-close-session; og-add-position -> og-add-seat; verb-object naming rule on two axes, consequence and frequency; under Director direction 2026-08-24)",
      "og-strategist <og-strategist@ogframework.orgdef> (v0.3.0 — communication substrate / seat notification split ratified: substrate declaration, ninth canonical pattern, Reserved Conventions bot identity to .orgdef addressing, OG's bespoke notification implementation retired from the forward surface; first charter entry authored under the new .orgdef identity; under Director direction 2026-08-24)",
      "og-strategist <og-strategist@ogframework.orgdef> (v0.3.1 — defect fix: the two current-state incumbent identifiers still carried @ogframework.com after v0.3.0 unified seat identity on .orgdef; surfaced by a fresh og-implementer orient. Under Director direction 2026-08-24)",
      "og-strategist <og-strategist@ogframework.orgdef> (v0.3.2 — relayed authority and directly-imposed constraints ratified: the two-class rule recorded in governance_model, with quote-don't-assert and hold-with-discharge-conditions; requested by the og-implementer seat after it held a build on a relayed lift. Under Director direction 2026-08-25)"
    ],
    "license": "MIT",
    "created": "2026-05-23",
    "extracted_from": "Bootstrap via PO direction 2026-05-23 in orgdef-strategist session, post data-vs-pattern sharpening. Inputs: thingalog-strategist hand-off memos at orgdef-spec/memos/ (2026-05-22-1430 pattern_promotion; 2026-05-22-1530 validation closeout; 2026-05-23-0900 hand-off; 2026-05-23-1100 addendum with data-vs-pattern sharpening). Reference shape: github.com/orgdef-spec/orgdef/orgdef-spec-organization.opencatalog (v2.0.0; sibling -def-spec org). Canonical template: github.com/orgdef-spec/orgdef/proposed-orgs/oagp-family-open-standard.opencatalog v2.0.0. Authoring discipline: orgdef-strategist authored the format-shape (artifact validity, structural placement, position roster shape); pattern-shape content (mission/values/red-lines/v1_criterion) sourced from the data-vs-pattern sharpening memo + canonical template inheritance, flagged for Director ratification.",
    "homepage": "https://ogframework.com (forthcoming)",
    "repository": "https://github.com/ogframework/og/blob/main/org/oagp-organization.opencatalog",
    "derived_from": {
      "id": "oagp-family-open-standard",
      "version": "2.0.0",
      "url": "https://github.com/orgdef-spec/orgdef/blob/main/proposed-orgs/oagp-family-open-standard.opencatalog"
    },
    "history": [
      {
        "version": "0.1.0",
        "date": "2026-05-23",
        "change": "Initial scaffolding via orgdef-strategist on PO direction post data-vs-pattern sharpening. v0.1.0 (not v1.0.0) because: (a) oagp-strategist seat is vacant — pattern-shape content awaits ratification by the staffed strategist seat; (b) embedded roledef:Job items deferred until seats staff; (c) recommended_patterns.general placeholder pending oagp-strategist pattern-promotion calls; (d) canonical pattern documentation at oagp.org does not yet exist. Promotion to v1.0.0 expected when v1_success_criterion items (b)–(d) are satisfied."
      },
      {
        "version": "0.1.1",
        "date": "2026-05-23",
        "change": "oagp-strategist seat staffed under Director direction during onboarding session (Claude Opus 4.7 at s:/projects/oagp-org). strategist.status flipped vacant→staffed; incumbent record added. First strategist decision drafted in-session: canonical promotion of the /oagp-bootstrap + /oagp-onboard skill pair (artifacts at proposals/oagp-adoption-cycle-canonical-promotion.md + decisions/proposal-oagp-adoption-cycle-canonical-promotion.md). Eight-item inherited work queue re-assigned from Director-interim to oagp-strategist; item 1 (canonical promotion) resolved by this version pending merge-ratification, items 2–8 remain active. Job specialization (embedded roledef:Job item) and recommended_patterns.general curation deferred to subsequent versions."
      },
      {
        "version": "0.1.2",
        "date": "2026-05-24",
        "change": "Canonical adoption-cycle skill distribution v0.1 + canonical promotion of /oagp-closeout drafted under Director direction 2026-05-24. Triggers: PO direction ('tying to thingalog makes no sense' and 'add one more skill first: oagp-closeout'); PO friction signal (memos/2026-05-24-0905); cross-spec residence coordination from orgdef-strategist (memos/2026-05-24-0900). Artifacts: proposals/canonical-adoption-cycle-skill-distribution-v0.1.md + decisions/proposal-canonical-adoption-cycle-skill-distribution-v0.1.md; proposals/oagp-closeout-canonical-promotion.md + decisions/proposal-oagp-closeout-canonical-promotion.md. Build: (a) skill content migrated verbatim from thingalog to oagp-org/skills/oagp-{bootstrap,onboard}/SKILL.md; (b) oagp-org/skills/oagp-closeout/SKILL.md newly authored as canonical session-cycle closing skill (companion to /oagp-onboard); (c) install scripts at install/install-claude-code-skills.{ps1,sh} cover all three skills; (d) README updated with Quick install (Claude Code) section and Canonical skills section reframed to list three skills with cycle-role labels; (e) cross-spec coordination memo filed to thingalog-strategist (memos/2026-05-24-1200) proposing thingalog/skills/ deprecation timeline. Distribution decisions versioned: v0.1 (Claude Code install, this); v0.2 (substrate-neutral primer URL, scheduled); v0.3 (Claude Code plugin packaging, scheduled after v0.2 per cross-vendor red line). Queue items 2 (transcript-tagging convention) resolved by closeout canonicalization for canonical case; 6 (plugin packaging) and 7 (cross-runtime delivery) partially addressed; bind/agent-sdk graduation (memos/2026-05-23-1600), canonical-orgs library residence reply to orgdef-strategist, and items 3–5 remain open."
      },
      {
        "version": "0.1.4",
        "date": "2026-05-24",
        "change": "oagp-implementer seat staffed under Director direction 2026-05-24 in fresh /oagp-onboard session (Claude Opus 4.7 1M context at s:/projects/oagp-org). implementer.status flipped vacant -> staffed; incumbent record added. Precondition: day-zero inbox memo at memos/2026-05-24-2000 enumerated the forward-work queue. Acceptance memo at memos/2026-05-24-2100 records the staffing event + first-move plan (clone oagp-org/oagp.org -> read current state -> report concretely -> propose specific content updates for PO ratification before substantive content work). Embedded roledef:Job item deferred per same precedent as oagp-strategist seat. CLAUDE.md roles-table patch drafted reflecting implementer staffed; broader status-line drift (v0.1.2 + v0.1.3 not folded into the narrative status block) flagged in acceptance memo §6 for oagp-strategist seat to fold in on a future pass. Director ratifies by merge per bounded-authority discipline; no push from implementer seat. Queue items 6 (plugin packaging), 7 (cross-runtime delivery), 8 (oagp.org canonical hosting) and implementer-seat-staffing forward-work item now have an active executor; v0.3 (Claude Code plugin) remains scheduled as last per cross-vendor red line ordering."
      },
      {
        "version": "0.1.3",
        "date": "2026-05-24",
        "change": "GitHub-org consolidation under oagp-org + canonical adoption-cycle skill distribution v0.2 drafted under Director direction 2026-05-24. Triggers: PO housekeeping direction ('I have a git organization called oagp-org... should we make it all available via MCP as well?'); sharpened substrate model from Director ('catdef is truly different. It's just a data spec. roledef, orgdef, memodef are all part of OAGP. Transcripts are a subcategory of memos.'); 2026-05-21 architectural lean from memodef-strategist on family-level MCP. (a) Three OAGP-internal -def repos moved under oagp-org GitHub org (oagp-org/{roledef, orgdef, memodef}); catdef-spec/catdef stays standalone as substrate-agnostic spec; openmemo-spec/openmemo-spec archived (historical dead-end with one commit). (b) oagp-online/oagp-online transferred and renamed to oagp-org/oagp.org (site repo + future MCP server host). (c) Canonical adoption-cycle skill distribution v0.2 drafted: ratifies memodef-strategist's 2026-05-21 architectural lean on family-level MCP at oagp.org/mcp (single MCP, per-spec namespaced tools, read-only convenience cache, repos remain canonical, per-spec strategist authority over namespaces); plus static primer.md fallback for non-MCP runtimes. (d) Six cross-spec coordination memos filed: 2026-05-24-1800 to memodef-strategist (move FYI + transcripts-as-memodef:Transcript confirmation), 2026-05-24-1801 to orgdef-strategist (move FYI), 2026-05-24-1802 to roledef-strategist (move FYI), 2026-05-24-1803 to catdef-strategist (catdef stays standalone FYI), 2026-05-24-1900 to memodef-strategist (MCP architectural lean ratified). (e) Internal-URL cleanup in operational content (5 files; historical artifacts preserved). Queue items 6 (plugin packaging) and 7 (cross-runtime delivery) substantially addressed by v0.2; v0.3 (Claude Code plugin) remains scheduled as last per cross-vendor red line ordering. Implementer-seat staffing now the natural next strategic question."
      },
      {
        "version": "0.1.5",
        "date": "2026-05-24",
        "change": "Canonical promotion of /oagp-init (founding-side skill) drafted under Director direction 2026-05-24: 'We need another skill. oagp-create_new_org... Technically this does not need git, but it does need a folder.' Artifacts: proposals/oagp-init-canonical-promotion.md + decisions/proposal-oagp-init-canonical-promotion.md; skills/oagp-init/SKILL.md. /oagp-init is the founding-side companion to /oagp-bootstrap (conversion-side); five-phase shape mirrors /oagp-bootstrap with Phase 1 = Elicit (interview-driven) instead of Survey (artifact-driven). Folder-only by default per Director direction ('the tech-savvy users will choose a repo. The rest will just choose a folder, and that's OK.'); git initialization is opt-in at Phase 4h. Sub-org case deferred to forward-reference idea memo per Director direction ('Agreed, parameter, but I don't think that's MVP. Now, you should absolutely file that as an idea'); memos/2026-05-24-2200 captures the non-trivial governance questions (sub-org Director authority scope, write-limits to subtree, merge gating, cross-subtree visibility, parent-org override, hierarchical-vs-flat OAGP question). Coordination memo to oagp-implementer (memos/2026-05-24-2201) for install-script mechanical update to add /oagp-init to skills array; small interleavable item within implementer's queue. Naming chosen as /oagp-init (single-verb-hyphen-cased; matches git init semantics; rejected /oagp-create-new-org, /oagp-new, /oagp-found per proposal alternatives §5). Init-helper transcript-position-tag convention ratified canonically (paralleling bootstrap-helper). CLAUDE.md known-work-items updated; charter authors list extended."
      },
      {
        "version": "0.1.6",
        "date": "2026-05-24",
        "change": "bind() prototype graduated to oagp-org/agent-sdk/ as v0.1 under Director direction 2026-05-24 ('P2 is substantially more important... let's go right to prototype'). Package: hatchling-built Python module under src/oagp_agent_sdk/ with bind() + BindResult dataclass; examples/bind_time_travel.py; pytest suite (36 unit + integration tests passing; no network). One API tightening from prototype: roledef_source str -> Literal['url','embedded']. Empirical lessons from the 2026-05-23 thingalog PoC preserved as defaults (acceptEdits permission_mode; flat path + single-hyphen names; URL-first roledef resolution with embedded fallback; color map per roledef id). Ratification request to oagp-strategist (memos/2026-05-25-0000); cross-spec coordination memos to roledef-strategist (URL-resolution contract, memos/2026-05-25-0001) and memodef-strategist (filename-timestamp sidebar, memos/2026-05-25-0002) filed in parallel. Entry drafted by implementer in memos/2026-05-25-0000 §4, ratified by oagp-strategist in memos/2026-05-28-1000, folded into the charter on the 2026-05-29 reconciliation pass."
      },
      {
        "version": "0.1.7",
        "date": "2026-05-28",
        "change": "bind() v0.1 ratified (interactive scope; memos/2026-05-28-1000): all seven §7 dispositions accepted, two open questions resolved; code review found bounded authority is conventional-not-structural in v0.1 (correct for interactive scope). agent-sdk v0.2 autonomous-dispatch direction drafted (proposals/agent-sdk-v0.2-autonomous-dispatch.md + decisions/): three-tier bounded-authority model -- (1) propose-only by construction (default); (2) explicit + audited Director elevation (overridable); (3) non-delegable dispatch (hard floor, no override, environmentally enforced). Per PO crystallization 'gun / foot / gun can never give instructions to another gun.' Bind-event memo discipline resolved; fail-closed roledef resolution gated on roledef-strategist contract (Option A). Folded into the charter on the 2026-05-29 reconciliation pass."
      },
      {
        "version": "0.1.8",
        "date": "2026-05-29",
        "change": "agent-sdk v0.2 amendment + OAGP layer-positioning (proposals/agent-sdk-v0.2-amendment-workflows-composition-and-positioning.md + decisions/), answering PO question whether native Claude Code Workflows makes bind() redundant. Answer: no -- amend, don't abandon. Evidence base: adversarial-analysis workflow wf_6a87be32-ee4 (9 agents; 4 lenses -> adversarial verify -> synthesis), treated as INPUT not authority per bounded-authority discipline. Crux: Workflows-dispatch maximizes task fan-out; OAGP-dispatch caps it (Tier-3) -- opposite axes; they compose (Workflows = within-session execution backend; OAGP = cross-session governance layer over it). Re-scope: run_seat() becomes the runtime-neutral abstraction composing over the native Workflows dispatcher as its Claude-Code backend (stop building bespoke dispatch plumbing); unique core untouched (roledef->AgentDefinition translation, time-travel binding, bind-event memos, three-tier authority). Tier-3 tightened to cover the Workflows-inside-autonomously-dispatched-bound-agent hazard. Layer-positioning language + fourth 'OAGP IS NOT' negation queued for README/primer. Pattern-promotion queue now 6 candidates; recommended_patterns.general remains placeholder pending the consolidated decision. Adoption-traction (v1 criterion d) escalated to Director, who deprioritized public launch 2026-05-29 pending further refinement. This reconciliation pass folds in the previously-uncommitted v0.1.6 + v0.1.7 + v0.1.8 charter entries (charter had drifted to v0.1.5 on disk while these decisions landed on main)."
      },
      {
        "version": "0.1.9",
        "date": "2026-05-29",
        "change": "Consolidated pattern-promotion v1 under Director direction 2026-05-29 ('Let's do the consolidated pattern-promotion'). Artifacts: proposals/consolidated-pattern-promotion-v1.md + decisions/proposal-consolidated-pattern-promotion-v1.md. recommended_patterns.general populated with eight canonical patterns, retiring the [DEFERRED] placeholder that stood since v0.1.0: (1) bounded-authority discipline; (2) seat-vs-incumbent persistence (absorbs the realized core of async-organization positioning); (3) data-vs-pattern layering; (4) org-state-fork-for-time-travel (validated x2); (5) substrate-is-sufficient-agent-context (validated by bind demos); (6) bounded autonomous dispatch / three-tier model (validated by the 2026-05-29 §8 demo; Tier-3 independently re-derived by Workflows); (7) org-governance-layer-above-runtime (consolidates the AGT + ETCLOVG + Workflows positioning; triply corroborated); (8) promotion-follows-adoption. Two candidates HELD pending more derivation (thingalog AI-PM/synthesis-agent seat, ships ~1-2 months; three-tier permission composition, 2 derivations + arguably substrate-MCP-surface-shape). Two DROPPED (canonical-by-reference holding-venue, superseded; bootstrap-helper transcript-tagging, already resolved in /oagp-closeout). Items kept as strings (array shape unchanged) -- a structured object shape would be orgdef format-shape, routed to orgdef-strategist if wanted, not taken here. Each promotion cleared the promotion-follows-adoption bar (independent re-derivation and/or empirical validation). This is v1 of the curated set, not the final word; future patterns promote via the same discipline as they earn it."
      },
      {
        "version": "0.2.0",
        "date": "2026-06-01",
        "change": "Rebrand OAGP -> OG (Open Governance Framework) under Director direction 2026-06-01; ogframework.com acquired (decisions/proposal-og-rebrand-open-governance-framework.md). Masthead identity rebranded OG-native: name -> 'OG — Open Governance Framework'; mission, vision, scope, the three values, and the four red_lines rewritten (OAGP->OG; 'pattern'->'framework' where it names the project, keeping the data-vs-pattern distinction as working vocab); homepage -> ogframework.com. Brand mark OG = 'Open Governance' (deliberately does not hard-code 'Agentic'). CURRENT SCOPE STAYS AGENTIC -- broadening to human organizations is a forward vision requiring separate ratification, explicitly NOT enacted here. Skill suite redesigned to seven og- skills (decisions/proposal-og-skill-architecture-v1.md): og-adopt, og-create, og-orient, og-claim-seat, og-closeout, og-snapshot, og-add-position. DEFERRED as a coordinated follow-on (these all touch the oagp-* identifier surface + the @oagp.org domain, and the seat-identifier rebrand changes git-author identities, so they migrate together rather than piecemeal): the seat-intertwined prose (governance_model, v1_success_criterion, position names/descriptions, recommended_roles, relationship descriptions); the seat identifiers themselves (oagp-strategist -> og-strategist; @oagp.org -> @ogframework.com); the orgdef id (oagp-org); the repository URL; and the GitHub repo/org rename. Until that follow-on those fields intentionally retain 'OAGP'/'oagp-*'/oagp.org (legacy redirect) -- the OG/oagp-* mix is a documented in-progress migration state, not an inconsistency. history[] and authors[] entries preserve 'OAGP' as-of-their-time (no retro-rebrand)."
      },
      {
        "version": "0.2.1",
        "date": "2026-06-01",
        "change": "Completed the OAGP->OG rebrand's coordinated identity+infra follow-on under Director direction 2026-06-01. GitHub: org oagp-org renamed to ogframework (UI action; sibling repos roledef/orgdef/memodef/oagp.org moved with it via redirect); main repo oagp renamed to og -> canonical github.com/ogframework/og; local remote updated. Charter: id oagp-org -> ogframework; repository URL updated; the previously-deferred seat-intertwined prose rebranded (governance_model, v1_success_criterion, recommended_patterns OAGP tokens, recommended_roles role ids, relationship descriptions, position names + descriptions); seat identifiers rebranded to the new canonical identities (og-strategist <og-strategist@ogframework.com>, og-implementer <og-implementer@ogframework.com>) -- the new git-author/bot identity going forward (this v0.2.1 commit is the first authored under it). PRESERVED as-of-their-time (no retro-rebrand): history[] entries, authors[] entries, and each position's staffed_via narrative (records of past events that used /oagp-onboard and oagp.org). CLAUDE.md + skills/install/README repo-URL + oagp.org references swept to OG/ogframework.com in the same pass. Cross-spec FYI memos filed to roledef/orgdef/memodef/catdef strategists re: the org rename. Remaining (not this repo's forward content): ogframework.com DNS/hosting (Director/ops); site+primer rebrand (implementer); sibling -def repos' own internal rebrand (each spec's strategist). The rebrand is complete in this repo's forward content."
      },
      {
        "version": "0.2.2",
        "date": "2026-08-24",
        "change": "Skill suite v2 ratified (decisions/proposal-og-skill-suite-v2.md) after a Director-led card-by-card review, prompted by an unbriefed field report from adopter org edgreenlight whose OG skills had vanished with a moved clone. Suite goes from seven cards to six: og-adopt + og-create merged into og-create-org (their SKILL.md phases 2-5 were identical; only Phase 1 differed, survey versus interview, and the merged Phase 1 does both -- og-adopt never instructed the seat to interview at all); og-snapshot -> og-describe-org (snapshot sat on the natural name of org-state-fork, pattern #4, a consequential operation, inside a suite organized around the safe-vs-consequential split); og-closeout -> og-close-session; og-add-position -> og-add-seat (pattern #2 is named seat-vs-incumbent, so the charter's own vocabulary settled the noun). Naming rule recorded as NEW rather than rediscovered, since the v1 suite did not follow it: verb + object, where the object is named whenever it is not obvious from the verb, plus a Director-contributed frequency corollary -- brevity earns its keep on a frequently-typed card, while a rare high-consequence card should be unmissable instead. Structural changes outweigh the renames: seams composed (describe-org <- orient <- claim-seat) so prerequisites are structural rather than prose; operating discipline moved into the generated CLAUDE.md so guardrails survive the skills going missing; the permanently-empty skills/ directory no longer created. Old names retained as aliases -- a requirement of acceptance, and the renames MUST NOT ship before the alias mechanism exists. Five candidate changes examined and rejected on the merits; og-describe-seat parked on one derivation."
      },
      {
        "version": "0.3.0",
        "date": "2026-08-24",
        "change": "Communication substrate and seat notification ratified (decisions/proposal-communication-substrate-and-seat-notification-v1.md), superseding but NOT withdrawing decisions/proposal-seat-inbox-notification-doorbell.md. Director-originated split of the durable COMMUNICATION SUBSTRATE from ephemeral SEAT NOTIFICATION, resolving a standing contradiction between value #3 (substrate-agnosticism) and CLAUDE.md's mandate that all cross-seat communication go through memos/ -- a substrate commitment hiding inside a pattern rule, surfaced by a real adopter moving to Jira. Every OG org ratifies one primary substrate (memos/ is the default) meeting four normative properties: durable and exportable, seat-addressable, auditable, enumerable. Ratification selects among conforming substrates; it does not confer conformance. OG now SHIPS NO NOTIFICATION COMPONENT -- orgs use whatever their runtime provides, and where a runtime provides none the substrate's enumerability is the answer, since a seat reads its inbox at session start anyway. OG's bespoke implementation (channels watcher, Claude Code channel adapter, vendor-neutral poll transport, attestation mechanism) is retired from the forward surface; it is NOT erased, and it is precisely what proved the convention is not vendor-shaped before being retired. Ninth canonical pattern added: seat-inbox-notification, promoted on evidence from three independent mechanisms carrying the semantic unchanged. Seat addressing adopted: <seat>@<charter-id>.orgdef, conferred by /og-claim-seat and never self-assigned -- the address is stable and org-scoped while the binding is disposable and runtime-specific -- superseding an earlier draft that made the runtime session name the address, which conflated the two and proved unreliable in practice. Git authorship unifies on the same identity (second seat-identity change since 2026-06-01, recorded as defect-driven rather than churn: @ogframework.com is a real-looking mailbox that is not one); historical commits keep their as-authored identities, no retro-rebrand. This org declares memos/ primary (record of strategy) and the OGF Jira project secondary (record of tactics). NOTE ON FORMAT-SHAPE RESTRAINT: the substrate declaration is expressed in an existing orgdef field (governance_model) rather than in a new top-level field, because adding a field to an orgdef artifact is format-shape and belongs to orgdef-strategist; a coordination memo proposing a first-class field is forward work. Cross-spec: memos/2026-08-24-1858 filed to orgdef-strategist on minting .orgdef as a pseudo-TLD (reply outstanding, non-blocking); a memo is owed to memodef-strategist on required communication-artifact semantics versus their encoding."
      },
      {
        "version": "0.3.1",
        "date": "2026-08-24",
        "change": "Defect fix in v0.3.0, surfaced within the hour by a fresh og-implementer session running /og-orient against the just-pushed substrate. v0.3.0 unified seat identity on <seat>@<charter-id>.orgdef and updated CLAUDE.md Reserved Conventions accordingly, but left both items[].incumbent.identifier fields on the superseded @ogframework.com address -- so the charter that ratified the addressing convention did not itself follow it. Corrected: strategist and implementer incumbent identifiers now og-strategist@ogframework.orgdef and og-implementer@ogframework.orgdef. The distinction applied: incumbent.identifier is CURRENT STATE and must track the ratified convention, whereas staffed_via narrative, authors[] and history[] are DATED RECORDS and keep their as-of-their-time identities (no retro-rebrand) -- which is why lines carrying @ogframework.com and @oagp.org elsewhere in this file are correct and were deliberately left alone. Recorded as its own version rather than silently patched into v0.3.0, since v0.3.0 was already pushed. Also surfaced by the same orient and NOT fixed here, because it needs a ratified convention rather than an edit: 26 memos carry action_required true, nearly all long discharged, so the substrate can enumerate ever-actioned items but not OPEN ones -- a defect against the enumerable property that v0.3.0 itself made normative (A2.4). Proposal to follow."
      },
      {
        "version": "0.3.2",
        "date": "2026-08-25",
        "change": "Relayed authority and directly-imposed constraints ratified (decisions/proposal-relayed-authority-and-direct-constraints-v1.md), answering a governance question raised by the og-implementer seat on 2026-08-24 after it HELD an authorised build rather than start on a relayed lift it could not verify. The rule: a seat may act on direction from a seat holding delegated authority over that work within an already-ratified shape, so prioritisation among ratified queue items needs no Director confirmation; but a seat may NOT act on a relayed lift, grant or exception where the relayed authority is not the relaying seat's to exercise -- notably a constraint the Director placed on a seat directly. THE TEST IS WHO IMPOSED THE CONSTRAINT, not how the instruction was worded: if the relaying seat could have issued it on its own authority it is delegated direction, otherwise it is reported authorisation. Obligations added: the relaying seat MUST quote the authoriser verbatim with context, MUST name any inference, and MUST NOT press the receiving seat's judgement; the receiving seat MUST hold absent a quote and MUST state what would discharge the hold, since a hold is not a refusal. Rationale: without the rule the org drifts into EROSION (seats normalising 'the strategist says the Director said so') or OSSIFICATION (seats blocking on the Director for work already ratified). Quoting is the remedy; the class distinction is the reason, since a rule saying only 'always quote' is discarded the first time it feels bureaucratic. Recorded in governance_model rather than as a recommended_patterns.general entry, following the v0.3.0 precedent of using an existing orgdef field rather than minting one -- and because promotion is held under pattern #8 on one derivation, one org, one incident. NOTE: the ratified proposal contained a direct contradiction -- its Target-version line claimed it added a recommended_patterns entry while its own Open Question 3 said it was not proposed for promotion. Resolved in favour of OQ3 as the considered position, and recorded in the decision rather than quietly fixed, since the Director ratified the document containing it. CLAUDE.md bounded-authority section extended in the same pass. The rule's first live test was its own ratification: the strategist declined to treat 'Sure, that looks fine' as ratification of a governance rule and asked for the Director's explicit word before triggering a charter bump."
      }
    ]
  }
}