[Tranche] Organization Governance Control Plane v0

[mdheller]

Purpose

Capture the Wand/agentic-labor benchmark into our estate as a concrete SocioProphet product tranche.

Wand validates the buyer category: enterprises want a governed control plane for human-agent work, not a loose pile of agents. Our counter-position is stronger and more explicit:

Wand manages agentic labor. SocioProphet proves, governs, audits, composes, and improves cybernetic institutions.

Canonical product loop

Objective → Workroom → Actor → Role → Policy → Asset → Action → Evidence → Review → Outcome → Score → Learning

This is the buyer-visible loop. The implementation should map across Ontogenesis, Agent Registry, Policy Fabric, AgentPlane, Model Governance Ledger, Sociosphere, Delivery Excellence, Sherlock, Prophet Workspace, and SourceOS.

Platform scope

prophet-platform owns the runtime/deployment hub for this tranche:

• platform-facing contracts for objectives, work orders, actor bindings, authority checks, evidence refs, review outcomes, and score/adoption telemetry;
• service/API bootstrap for the OrgGov control loop;
• smoke-test wiring for a dogfood GitHub issue → policy gate → agent/human action → evidence → review → scorecard workflow;
• integration with existing evidence, receipt, TriTRPC, evaluation fabric, and runtime surfaces;
• docs that make this a single product slice instead of scattered repo doctrine.

Required child workstreams

• prophet-workspace: control-room UX and Professional Workroom binding.
• ontogenesis: canonical organizational work ontology.
• agent-registry: actor, role, authority, tool grant, revocation bindings.
• policy-fabric: role/action/asset policy gates and replay packs.
• agentplane: work-order-to-run-artifact and replay binding.
• model-governance-ledger: model/tool/data receipt linkage.
• sociosphere: estate topology, canonical ownership, dependency propagation.
• delivery-excellence: KPI/OKR scorecard and adoption telemetry.
• sherlock-search: searchable evidence/control graph.
• sourceos-syncd: local-first state integrity and repair/evidence integration.

Acceptance criteria for v0

• A canonical orggov contract set exists under contracts/ or an equivalent platform-owned path.
• One demo workflow can be executed or simulated end-to-end.
• The workflow emits evidence references and adoption/score telemetry.
• Policy gates run before action execution or output release.
• The contract names match the ontology and registry semantics.
• Workroom, actor, role, policy, asset, evidence, outcome, and score objects are visible in a product-readable model.
• CI/smoke checks verify schema validity and at least one golden fixture.

Non-goals

• Do not clone Wand’s closed product language.
• Do not build a generic task manager.
• Do not bypass evidence for demo speed.
• Do not make chat the product. Chat is only one control surface over governed institutional work.

Strategic note

We already have most of the machinery. This tranche compresses the machinery into a coherent product body: a visible control plane for governed human-agent institutions.

[mdheller]

OrgGov v0 platform contract spine is now opened as PR #409.

PR: #409

Captured artifacts:

• orggov.manifest.yaml
• contracts/orggov/README.md
• contracts/orggov/orggov-control-plane.v0.1.schema.json
• contracts/orggov/orggov-control-plane.v0.1.example.json
• docs/ORG_GOVERNANCE_CONTROL_PLANE_V0.md
• tools/validate_orggov_contracts.py

Local validation result:

ok: manifest references 2 existing contract paths
OK: Organization Governance Control Plane validation passed

This PR establishes the platform-owned spine for the control loop:

Objective → Workroom → Actor → Role → Policy → Asset → Action → Evidence → Review → Outcome → Score → Learning

Follow-on implementation remains split across the child workstream issues created for Ontogenesis, Prophet Workspace, Agent Registry, Policy Fabric, AgentPlane, Model Governance Ledger, Sociosphere, Delivery Excellence, Sherlock Search, and SourceOS syncd.