OBE v1.0 must be built in an order that locks architecture before runtime behavior. Do not jump directly to Auto Post implementation before the AI, Library, Safety, and Activity contracts are ready.
The historical blueprint sequence remains below, but current mission tracking now lives in Project Command Center, Project Status, Milestones, and Next Missions. As of v0.19.0, the Editorial Humanizer / Readability addon foundation is complete as a bundled Community addon. The next runtime target is v0.19.0 hardening while preserving Library-first, review-first behavior.
- Master Architecture
- UI Blueprint
- AI / Responses API Contract
- AI / Responses API Implementation
- Library
- Activity Log
- Dry-run + Approval
- Auto Post Source Preview
- Field Mapping
- Auto Post Planner
- AI Draft Generator
- Draft Writer
- Addon registry and module boundary
- Workflow Pro / Agents SDK
- Final hardening
Lock the public OBE identity as Optimization-Based Engine and define layers, boundaries, safety invariants, addon placement, and build order.
Design WordPress-native page layouts, widths, cards, tables, badges, forms, detail pages, empty states, and approval confirmations.
Define AI_Request, AI_Response, model profiles, task types, structured outputs, redaction, usage tracking, Provider Interface, Library integration, Activity integration, and Responses API payload mapping. This is the contract step before implementation and must include no runtime PHP or live API calls. OpenAI defaults to Responses API.
Next implementation mission after the contract PR: feat: add AI request/response contracts. Implement the AI request/response classes and then the OpenAI provider wrapper after the contract is stable. Do not use Chat Completions as the default OpenAI integration.
Implement private CPT-backed Library items and statuses so generated work is reviewable before writing.
Record provider settings, key save/clear, source preview, mapping, validation, Library creation, AI generation, approval, rejection, and draft write events with redaction.
Implement dry-run result models, approval states, write gates, and confirmation behavior.
Add pasted text, CSV, XLSX/export, and generic Partner API Source preview concepts without private adapters.
Map generic fields only. Do not include private field maps or client-specific business rules.
Create plans in Library before any draft generation or writing.
Generate content_draft Library items using the AI Engine and selected provider.
Write only approved Library content_draft items to draft or needs_review. Publish only after separate explicit approval.
Add SEO Audit & Fix, Performance Audit, Auto Content / Auto Post, Translation, and Workflow Pro cards under OBE → Addons. Addons must not create separate top-level menus.
Keep Agents SDK out of Community Core. Consider it later for Workflow Pro or a separate service when OBE owns orchestration, tool execution, approvals, state, custom storage, and runtime behavior.
Perform full security and reliability hardening: capability audit, nonce audit, SQL audit, REST permission audit, file upload audit, log redaction audit, rollback verification, rate limiting, dependency review, and compatibility checks.
Final hardening happens late, but every implementation stage must include capability checks, nonce checks for writes, sanitization, escaping, masked keys, no raw secret display, no direct publish, and no destructive default.
OBE Core owns the admin shell, provider settings, AI engine contracts/orchestration, Library, Activity, Approval, Safety, Writing, and Addon registry/contracts. Bundled Community addons own feature workflows such as Auto Post / Import, SEO Review, Performance Review, and Translation Plan. Future Pro/private addons, private adapters, private prompts, private field maps, partner API connectors, and production/client-specific workflows are not included in the public runtime.
v0.18.0— Editorial Humanizer / Readability Addon Foundation.