Changelog

Append-only log of notable changes to the portal and the ecosystem it maps. Newest first. Each material change (new repo, new venture, new diagram edge, new ADR) gets one line.

Format: YYYY-MM-DD — <what changed> (<affected pages / ADR>)

2026-07-06 — Publication pipeline implementation (Phase 0–3, autonomous sprint)

Implemented the frozen Publication Model as a zero-dependency, deterministic build under site/scripts/publication/ (contract lib · generators · validation · tests; npm run build:publication / validate:publication / test:publication). Generates 16 feeds + 14 JSON Schemas + manifest into publication/** (git-ignored build artefact, like graph/health).
Live/buildable (14): canonical spine, routes, navigation, reference-library (12 collections), portfolio- gallery (capabilities + systems), search-index, living-platform (health snapshot), architecture-gallery (Mermaid exhibits), engineering-manual, evidence-plates (sourced from the flagship digest), arrival (editorial hero migrated into content/editorial/), flagship-experiences (Website Assessment extracted from its narrative page). Awaiting (2, honest): intelligence-products (seed WA pack; pending intelproducts export), story (chapter shells; prose pending editorial migration).
Validation harness is green (schema · manifest integrity · routes · spine/nav/search single-source · classes · stats cross-check · render-safety · evidence · terminology); rebuilds are byte-identical (determinism test).
Contradiction fix: added structural page types (home/overview/registry/meta/doc) to the entityType enum so routes/reference can type every page (recorded in the contract doc). No other frozen contract changed.

2026-07-06 — Publication Model pre-freeze corrections + feed dependency map

Applied the Final Architecture Review’s agreed pre-freeze changes across all publication docs (design only): a Claude-owned Editorial Source Layer (content/editorial/**) so authored narrative is no longer owned by presentation; a reserved Live feed class (mutable/TTL) distinct from Static/Generated; System vs Venture split with ventures reserved first-class; a single Canonical Platform Spine (canonical/spine.json) referenced by every destination; navigation.json and search-index.json contracts; a scaling reservation (index/detail/shard/page); publication classes with class-routed approval (auto-publishable vs human-approved); and canonicalised naming (Flagship Experiences, Living Platform, System, Venture).
Added publication-feed-dependency-map.md — full repository→feed→portal traceability. Verdict: READY TO FREEZE.

2026-07-06 — Publication Model & Lovable Data Contract (design handoff)

Added the truth→presentation bridge design: Publication Model, Lovable Data Contract, Publication Pipeline, Entity Catalogue, Validation, and Implementation Roadmap.
Defines ownership (Claude = truth, Lovable = presentation), 9 destination feeds + core feeds, feed tiers (live/buildable/awaiting/future), evidence-first missing-data rules, the deterministic gated pipeline, and a contract-first roadmap. Design only — no feeds generated, no frontend built; FIP mechanics and intelligence- product feeds honestly labelled awaiting.

2026-07-06 — Flagship showcase: Website Assessment capability

Restructured Website Assessment from a documentation page into an evidence-based narrative showcase — the platform’s canonical demonstration. New flow: the problem → why existing approaches fail → how the platform approaches it → the end-to-end journey (now the primary visual) → architecture by layer → the output + real artefacts → business value → capability composition (why it’s reused, not rebuilt) → the decisions that shaped it → consolidated evidence & maturity (implemented / in progress / planned).
Content/communication only — no new architecture, no layout change, no maturity claims altered. All evidence reused from source (pack contract, API/COP contracts, assessment-engine-plan-v2, scoring-model, pack-manifest); live sample outputs explicitly noted as living in the repo, not fabricated here.

2026-07-06 — Phase G4.2: refined the executive Portfolio Overview

Reorganised the Portfolio Overview into two clear halves — The Timeless Architecture (problem · vision · architecture-in-one-picture · layered platform · information flow · Website Assessment example) and The Current State (repositories · maturity · technology · direction) — so the design is separated from the status and the page is easier to maintain.
Expanded the Vision to explain architectural reasoning (why reusable intelligence, why capabilities are separated from applications, why immutable versioned products over shared prompts, why shared services) and added “The architecture in one picture” (Knowledge → Capabilities → Intelligence → Products → Applications → Ventures → Customer outcomes) as the conceptual anchor.
Made Website Assessment the canonical recurring example; reduced implementation detail (technology, maturity, roadmap now summarised and linked out); strengthened the close on the compounding idea. Editorial only — no architecture, motivations, or maturity claims changed.
Moved the portal’s Phase C/D/E/F history out of the executive narrative into a new Portal Evolution page.

2026-07-05 — Phase G1: Portfolio Graph + website foundation (implementation)

Built the Portfolio Graph — a build-time, typed runtime model generated from the markdown (150 nodes · 739 typed edges: implements/consumes/belongsToLayer/inSystem/usesTechnology/relatesTo/…). Zero-dependency generator at site/scripts/build-graph.mjs; schema at site/GRAPH-SCHEMA.md; exposed at GET /api/graph.json. Data flow: Markdown → Portfolio Graph → Website → (future) Intelligence/APIs.
Built the initial custom Astro website (site/) consuming the graph: layouts, routing (70 pages, each generated from the graph), header/sidebar navigation, dark/light theme, client-side theme-aware Mermaid, relationship-aware navigation (typed relationship panels), entity-aware search, a platform homepage (hero · scale strip · interactive layers · capability & technology summaries · entry points), and Cloudflare Pages config. Build validated (npm run build → 70 pages, 0 leftover .md links).
ADR 0007 — Portfolio Graph as canonical runtime model; custom Astro (not Starlight).
Additive only: all markdown stays in place; site/ is fully regenerable; generated graph + build artefacts are git-ignored. Markdown remains the single source of truth.

2026-07-06 — Phase G3.1: closed high-priority Portfolio Health gaps

The deterministic Portfolio Health engine flagged three high-priority gaps (registered-but-undigested repos). Processed all three to the digest standard, evidence-first (cross-repo evidence hunt): inexisstudios (Website Assessment engine + Inexis Digital site/CMS), outreachagent (outbound cold-outreach execution engine), and leadplatform (inbound Upwork opportunity intelligence).
Added the Applications & Agents layer-4 system digest — the website-assessment delivery pipeline (inexisstudios → outreachagent → Inexis Digital; leadplatform inbound).
Fixed a graph slug-collision (the shared-skills repo vs the shared-skills system): capability provider/consumer edges now resolve to repositories, systems to systems — the shared-skills repo’s capability mapping is now correct (resolves one work-queue item). No graph-model change.
Updated the registry (all 7 repos now fully digested), repos/ventures indexes, portfolio overview map, and status dashboard. Graph + Portfolio Health regenerated — scores and coverage improved from real evidence.

2026-07-05 — Phase F: Architecture Governance & Decision Framework

Established the Architecture Governance section (synthesis of existing discipline; no new repos): objectives, decision process, ownership, review, approval workflow, quality gates, and change-management principles.
New pages: Capability Decision Framework (decision trees), Capability Decision Matrix (question → destination standard), Repository Lifecycle (proposal → archival), Documentation Governance (source of truth per artefact), AI Governance (authority boundaries + required human approvals), and the Architecture Review Lifecycle (triggers + checklist).
ADR 0006 — adopt the governance framework (a portal-governance decision; no ecosystem architecture changed).
Distinguished implemented governance (ADRs, phase-gated review, Definition of Done, human approval gates) from recommended future governance (annual review cadence, automated pre-review checks, orchestrator sync).

2026-07-05 — Phase E: Operating Model & Knowledge Architecture

Established the Operating Model section (synthesis of existing evidence; no new repos): the end-to-end work lifecycle with human / AI-assisted / automated / future-autonomous activity distinctions.
New pages: Knowledge Asset Lifecycle, Concept Ownership Registry (ambiguous ownership flagged), authoritative Portfolio Glossary, Context Architecture, Operating Principles (split from architecture principles), and the AI Collaboration Model.
Honest labelling: flagged concepts with ambiguous ownership (Recommendations, Opportunity Scores, Website Assessments, Maturity levels); recorded “Domain Briefs” and “Review Packages” as not-evidenced rather than inventing owners. Future autonomy (OpenClaw, Hermes, orchestrator) kept clearly separate from implemented.
Documentation only — no architecture modified. Committed Phases A–D as portal v1 (feat: establish portfolio architecture portal v1).

2026-07-05 — Phase D: Ecosystem Architecture Review

Performed a holistic, evidence-first EA review of the ecosystem (no new repos processed). New pages: Architecture Review (ownership, duplication, layer boundaries, dependency health, reuse, information flow), Architecture Risks (12-item register), Strategic Opportunities, and Recommended Architecture Evolution (current → recommended → vision).
Updated the Capability Reuse Map with a Phase-D reusability verdict.
ADR 0005 (proposed) — introduce a Platform Services layer + consolidate reusable services (single scoring source; extract the Assessment Service out of the venture repo). Left proposed — recommendations only, no architecture modified.
Headline findings: reusable capabilities trapped in venture repos (assessment engine in inexisstudios); duplicated opportunity scoring; unbuilt foundational loops; mostly-mock cross-repo consumption.

2026-07-05 — Phase C: Capability Architecture + Website Assessment

Introduced Capability Architecture — new top-level portal/capabilities/ section with a Capability Registry (primary nav entry point), a capability template, and 7 capability pages.
Processed the Website Assessment Platform as the first cross-repo platform capability, with 4 architecture views (executive, container, component, runtime). Evidence gathered directly from inexisstudios, the intelproducts pack, outreachagent, the lab, and FIP.
New Capability Reuse Map (providers, consumers, shared assets, intelligence dependencies, reuse opportunities, duplication risks — table + Mermaid).
ADR 0004 — introduce Capability Architecture. Added Architecture Principles 11–12 (deterministic rules decide / AI explains; capabilities are first-class).
Registered inexisstudios (the assessment engine, documented at capability level). Updated Portfolio Overview, registry, system dependency map, tech landscape (Playwright, TanStack Start, D1/R2), status dashboard, IA/nav, and llms.txt.
Honest labelling: engine core built (self-tests pass) but cross-repo consumption mostly mock; scoring weights are inference-derived; Hermes learning loops unimplemented; leadplatform assessment consumption declared-not-wired; flagged two parallel opportunity-scoring models as a duplication risk.

2026-07-05 — Phase B: Intelligence Platform layer

Processed the Intelligence Platform as a platform, not a single repo. Evidence gathered directly from source repos (with a read-only cross-repo evidence hunt).
New platform doc Intelligence Platform architecture with four views: executive overview, container, component, and runtime information flow.
New digests: PersonalOps/FIP (producer + Intelligence Dashboard, layer 2), website-intelligence-lab (engine evaluation, layer 2), intelproducts (published packs, layer 3). New system digest Intelligence Platform layer.
New ecosystem-wide Architecture Principles page (10 principles), referenced across the portal.
ADR 0003 — model the Intelligence Platform as a multi-system layer with a producer/consumer product contract.
Registered downstream consumers leadplatform + outreachagent (layer 4, not yet fully processed).
Updated Portfolio Overview (honest maturity, five ventures, evidenced tech landscape), system-of-systems diagrams + dependency map, status dashboard, IA/nav, and llms.txt.
Honest labelling: OpenClaw & Hermes recorded as planned/not-built; “estate generation” recorded as NOT FOUND (not a real subsystem); the recommendation engine noted as out-of-scope (inexisstudios).

2026-07-05 — Phase B: first real repository

Processed the Shared Skills repository (169 skills) as the first real repo; created the reference digest repos/shared-skills.md and the ventures/shared-skills.md system/layer digest. (Skills Layer doc)
Upgraded the repo-digest template to the richer standard the flagship digest establishes.
Added the top-level Portfolio Overview executive briefing (vision, layered architecture, information flow, portfolio map, maturity, tech landscape, strategic roadmap).
Adopted the six-layer ecosystem model and honest realization labelling. (see ADR 0002)
Populated architecture pages with layered diagrams and a System Dependency Map; added architecture review recommendations (5 duplication flags, reusable-capability and missing-doc findings).
Updated registry, status dashboard, indexes, IA/navigation, and llms.txt.
Removed the Phase A _example-repo / _example-venture placeholders now that real data has landed.

2026-07-05 — Phase A: initialization

Portal initialized — Phase A scaffold: information architecture, documentation model, templates, portal layer (registry, repos, ventures, architecture, current-state, decisions), AI handoff files, and the orchestrator skill spec. (see ADR 0001)