All Plan Forge architecture references in one place — from the unified system design to three-tier memory architecture and enterprise deployment blueprints.
Architecture reference for integrating Plan Forge, OpenBrain, and OpenClaw into a single automated development system — closed-loop from feature request to review with full memory continuity.
UNIFIED-SYSTEM-ARCHITECTURE.md → Enterprise · Appendix KOne canonical architecture for a 5-team / 1000-developer fleet, plus the Microsoft Foundry composition variant for Azure-tenant deployments. Includes three isolation patterns, capacity planning, and failure modes.
manual/enterprise-reference-architecture.html →Three tiers of memory organised by substrate — L1 in-process hub, L2 structured files, L3 OpenBrain semantic store — with write paths, lifetime guarantees, and guidance for MCP tool authors.
MEMORY-ARCHITECTURE.md → Chapter 24 · Act IVThree tiers, one capture path. How Plan Forge remembers what it learned — across slices, across sessions, across plans. Covers dual-write capture, watcher integration, drain queue, and migration tooling.
manual/memory-architecture.html →Semantic chunk index and BFS call-graph traversal for any git repository. Produces .forge/lattice/chunks.jsonl and edges.jsonl. Supports pure-JS chunker with optional tree-sitter upgrade. Five MCP tools: index, stat, query, callers, blast.
Lightweight, dependency-free provenance contract (hallmark/v1) shipped in pforge-sdk/hallmark. Validates, builds, and merges provenance records with a JSON Schema envelope. Two MCP tools: forge_hallmark_show and forge_hallmark_verify for drift detection.