hyperpolymath · hyperpolymath · Mar 13, 2026 · Mar 13, 2026
diff --git a/.claude/CLAUDE.md b/.claude/CLAUDE.md
@@ -11,73 +11,132 @@ The following files in `.machine_readable/` contain structured project metadata:
 
 ---
 
-# CLAUDE.md - AI Assistant Instructions
+# CLAUDE.md - Bofig Project Instructions
+
+## Current Phase: Phase 2 — Lithoglyph Migration
+
+**ADR-006 (2026-03-13):** Lithoglyph replaces ArangoDB as primary data store.
+
+### Pipeline
+```
+Docudactyl (extraction) → Lithoglyph (storage + provenance) ← bofig (queries + visualisation)
+```
+
+### Migration Status
+- Evidence reads/writes: **migrating** from ArangoDB → Lithoglyph GQL
+- Entities: **migrating** from ArangoDB → Lithoglyph
+- Claims: **pending** migration to Lithoglyph
+- Relationships (graph edges): **ArangoDB** (kept until Lithoglyph Factor GQL gets graph traversals)
+- User auth: **PostgreSQL** (permanent, phx.gen.auth)
+
+### Critical Architecture Rules
+1. **New domain data features MUST target Lithoglyph**, not ArangoDB
+2. **ArangoDB is deprecated for domain data** — only `relationships` edge collection remains
+3. **PostgreSQL is for user auth ONLY** — never store domain data there
+4. **PROMPT scores** have exactly 6 dimensions (Provenance, Replicability, Objective, Methodology, Publication, Transparency)
+5. **Audience types** are: researcher, policymaker, skeptic, activist, affected_person, journalist
+
+## Build & Test
+
+```bash
+# Compile (0 warnings required)
+mix compile --warnings-as-errors
+
+# Tests (require PostgreSQL + ArangoDB running)
+mix test                    # 257 tests
+
+# NER extractor tests only (no DB needed)
+MIX_ENV=test mix run --no-start -e '
+  ExUnit.start(autorun: false)
+  Code.require_file("test/evidence_graph/lithoglyph/ner_extractor_test.exs")
+  ExUnit.run()
+'
+
+# Credo lint
+mix credo --strict
+
+# Start dev server
+mix phx.server
+```
+
+## Key Modules
+
+| Module | Purpose | DB |
+|--------|---------|-----|
+| `EvidenceGraph.Lithoglyph.Client` | Req HTTP client for Lithoglyph API | Lithoglyph |
+| `EvidenceGraph.Lithoglyph.Importer` | GenServer batch import with NER | Lithoglyph → ArangoDB |
+| `EvidenceGraph.Lithoglyph.NERExtractor` | Regex NER extraction from content | None (pure) |
+| `EvidenceGraph.Entities` | Entity resolution, fuzzy match, merge | ArangoDB (migrating) |
+| `EvidenceGraph.Claims` | Claim CRUD + PROMPT scoring | ArangoDB (migrating) |
+| `EvidenceGraph.Evidence` | Evidence CRUD + metadata | ArangoDB (migrating) |
+| `EvidenceGraph.Relationships` | Graph edges, traversals, contradictions | ArangoDB (kept Phase 2) |
+| `EvidenceGraph.ArangoDB` | ArangoDB driver wrapper | ArangoDB |
+
+## ArangoDB Query Patterns
+
+```elixir
+# Read query (no write transaction)
+ArangoDB.query_read(aql, %{bind_var: value})
+
+# Write query (transactional)
+ArangoDB.query(aql, %{bind_var: value})
+
+# Insert document
+ArangoDB.insert("collection_name", %{field: value})
+
+# Edge document format
+%{
+  _from: "evidence/evidence_123",
+  _to: "entities/entity_456",
+  relationship_type: "mentions",
+  weight: 1.0,
+  confidence: 0.9
+}
+```
+
+## Lithoglyph Client Patterns
+
+```elixir
+# Query evidence from Lithoglyph
+LithClient.query("SELECT * FROM evidence WHERE investigation_id = @id", %{id: inv_id})
+
+# Insert with provenance (mandatory)
+LithClient.insert("evidence", document, actor: "user:123", rationale: "Import from Docudactyl")
+
+# Dedup check
+LithClient.exists_by_hash?("evidence", sha256_hash)
+```
 
 ## Language Policy (Hyperpolymath Standard)
 
 ### ALLOWED Languages & Tools
 
 | Language/Tool | Use Case | Notes |
 |---------------|----------|-------|
+| **Elixir** | This project's primary language | Phoenix, LiveView, Absinthe |
 | **ReScript** | Primary application code | Compiles to JS, type-safe |
 | **Deno** | Runtime & package management | Replaces Node/npm/bun |
 | **Rust** | Performance-critical, systems, WASM | Preferred for CLI tools |
-| **Tauri 2.0+** | Mobile apps (iOS/Android) | Rust backend + web UI |
-| **Dioxus** | Mobile apps (native UI) | Pure Rust, React-like |
 | **Gleam** | Backend services | Runs on BEAM or compiles to JS |
 | **Bash/POSIX Shell** | Scripts, automation | Keep minimal |
-| **JavaScript** | Only where ReScript cannot | MCP protocol glue, Deno APIs |
-| **Nickel** | Configuration language | For complex configs |
+| **JavaScript** | Only where ReScript cannot | D3.js hooks in this project |
 | **Guile Scheme** | State/meta files | STATE.scm, META.scm, ECOSYSTEM.scm |
-| **Julia** | Batch scripts, data processing | Per RSR |
-| **OCaml** | AffineScript compiler | Language-specific |
-| **Ada** | Safety-critical systems | Where required |
 
 ### BANNED - Do Not Use
 
 | Banned | Replacement |
 |--------|-------------|
 | TypeScript | ReScript |
 | Node.js | Deno |
-| npm | Deno |
-| Bun | Deno |
-| pnpm/yarn | Deno |
+| npm/Bun/pnpm/yarn | Deno |
 | Go | Rust |
 | Python | Julia/Rust/ReScript |
-| Java/Kotlin | Rust/Tauri/Dioxus |
-| Swift | Tauri/Dioxus |
-| React Native | Tauri/Dioxus |
-| Flutter/Dart | Tauri/Dioxus |
-
-### Mobile Development
-
-**No exceptions for Kotlin/Swift** - use Rust-first approach:
-
-1. **Tauri 2.0+** - Web UI (ReScript) + Rust backend, MIT/Apache-2.0
-2. **Dioxus** - Pure Rust native UI, MIT/Apache-2.0
-
-Both are FOSS with independent governance (no Big Tech).
-
-### Enforcement Rules
-
-1. **No new TypeScript files** - Convert existing TS to ReScript
-2. **No package.json for runtime deps** - Use deno.json imports
-3. **No node_modules in production** - Deno caches deps automatically
-4. **No Go code** - Use Rust instead
-5. **No Python anywhere** - Use Julia for data/batch, Rust for systems, ReScript for apps
-6. **No Kotlin/Swift for mobile** - Use Tauri 2.0+ or Dioxus
-
-### Package Management
-
-- **Primary**: Guix (guix.scm)
-- **Fallback**: Nix (flake.nix)
-- **JS deps**: Deno (deno.json imports)
+| Java/Kotlin | Rust |
 
 ### Security Requirements
 
 - No MD5/SHA1 for security (use SHA256+)
 - HTTPS only (no HTTP URLs)
 - No hardcoded secrets
-- SHA-pinned dependencies
-- SPDX license headers on all files
-
+- SHA-pinned dependencies in workflows
+- SPDX license headers on all files (`PMPL-1.0-or-later`)
diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS
@@ -0,0 +1,19 @@
+# SPDX-License-Identifier: PMPL-1.0-or-later
+# CODEOWNERS - Evidence Graph for Investigative Journalism (bofig)
+
+# Default owner for all files
+* @hyperpolymath
+
+# Core domain logic
+/lib/evidence_graph/ @hyperpolymath
+
+# Lithoglyph integration (Phase 2)
+/lib/evidence_graph/lithoglyph/ @hyperpolymath
+
+# Security-sensitive files
+/config/ @hyperpolymath
+/.github/workflows/ @hyperpolymath
+/SECURITY.md @hyperpolymath
+
+# Machine-readable state
+/.machine_readable/ @hyperpolymath
@@ -1,10 +1,10 @@
 ; SPDX-License-Identifier: PMPL-1.0-or-later
-; Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <jonathan.jewell@open.ac.uk>
+; Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
 
 (ecosystem
   (metadata
-    (version "1.0.0")
-    (last-updated "2026-02-21")
+    (version "1.1.0")
+    (last-updated "2026-03-13")
     (format "ECOSYSTEM.scm v1"))
 
   (project
@@ -17,18 +17,35 @@
   (position-in-ecosystem
     (domain "investigative-journalism-tools")
     (novelty "First system combining PROMPT scoring, boundary objects, and i-docs navigation")
-    (academic-context "PhD thesis: practical infrastructure for pragmatic epistemology"))
+    (academic-context "PhD thesis: practical infrastructure for pragmatic epistemology")
+    (pipeline-position "Docudactyl (extraction) -> Lithoglyph (storage+provenance) -> bofig (navigation+visualisation)"))
 
   (related-projects
-    (project "formdb-debugger" (relationship "sibling-standard"))
-    (project "formbase" (relationship "sibling-standard"))
-    (project "hypothesis" (relationship "inspiration") (url "https://hypothes.is/"))
-    (project "zotero" (relationship "integration-target") (url "https://www.zotero.org/")))
+    (project "lithoglyph"
+      (relationship "primary-dependency")
+      (role "Evidence store, provenance layer, GQL/GQL-DT query engine")
+      (notes "Phase 2: migrating all domain data from ArangoDB to Lithoglyph"))
+    (project "docudactyl"
+      (relationship "upstream-dependency")
+      (role "Multi-format HPC document extraction, feeds evidence into Lithoglyph"))
+    (project "verisimdb"
+      (relationship "sibling-standard")
+      (role "Octad database with VQL, shares GQL patterns with Lithoglyph"))
+    (project "zotero"
+      (relationship "integration-target")
+      (url "https://www.zotero.org/")
+      (role "Reference management, two-way sync for evidence import/export"))
+    (project "hypothesis"
+      (relationship "inspiration")
+      (url "https://hypothes.is/")
+      (role "Web annotation model")))
 
   (technology-stack
-    (runtime "BEAM/OTP 26+")
-    (language "Elixir 1.16+")
-    (framework "Phoenix 1.7+ with LiveView")
-    (database "ArangoDB 3.11+")
-    (api "Absinthe GraphQL")
+    (runtime "BEAM/OTP 27")
+    (language "Elixir 1.18+")
+    (framework "Phoenix 1.8+ with LiveView")
+    (evidence-store "Lithoglyph (GQL/GQL-DT, Phase 2+)")
+    (graph-db "ArangoDB 3.11+ (relationships only, Phase 2; removed Phase 3)")
+    (auth-db "PostgreSQL 16")
+    (api "Absinthe GraphQL + REST")
     (visualization "D3.js v7")))
@@ -1,18 +1,19 @@
 ; SPDX-License-Identifier: PMPL-1.0-or-later
-; Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <jonathan.jewell@open.ac.uk>
+; Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>
 
 (meta
   (metadata
-    (version "1.0.0")
-    (last-updated "2026-02-21")
+    (version "1.1.0")
+    (last-updated "2026-03-13")
     (format "META.scm v1"))
 
   (architecture-decisions
     (adr "001"
       (title "ArangoDB for primary data store")
-      (status "accepted")
+      (status "superseded")
+      (superseded-by "006")
       (decision "Use ArangoDB 3.11+ with Arangox Elixir driver via MintClient")
-      (rationale "Production-proven, multi-model (document + graph), managed hosting available"))
+      (rationale "Production-proven, multi-model (document + graph), managed hosting available. Superseded: Lithoglyph now provides these capabilities with additional provenance and type safety."))
 
     (adr "002"
       (title "Phoenix LiveView over React SPA")
@@ -36,9 +37,18 @@
       (title "Ecto without SQL for domain models")
       (status "accepted")
       (decision "Use Ecto schemas and changesets for validation only, not persistence")
-      (rationale "Leverage Ecto's validation without coupling to SQL")))
+      (rationale "Leverage Ecto's validation without coupling to SQL"))
+
+    (adr "006"
+      (title "Lithoglyph replaces ArangoDB as primary data store")
+      (status "accepted")
+      (date "2026-03-13")
+      (supersedes "001")
+      (decision "Migrate domain data (evidence, claims, entities) from ArangoDB to Lithoglyph. ArangoDB retained for graph edges only during Phase 2, fully removed in Phase 3.")
+      (rationale "Eliminates data duplication (Docudactyl→Lithoglyph→ArangoDB was a copy pipeline). Lithoglyph provides mandatory provenance, WAL audit trail, GQL-DT dependent types for PROMPT scores, and compile-time verification. ArangoDB was correct for Phase 1 prototyping but duplicates what Lithoglyph already stores.")))
 
   (design-rationale
     (principle "Infrastructure for pragmatic epistemology")
     (principle "Navigation over narration")
-    (principle "Coordination without consensus")))
+    (principle "Coordination without consensus")
+    (principle "Query the source, don't copy the data")))