docs: convert drafts into website posts

Author: AlexMikhalev

blog-posts/github-runner-architecture.md …blog-posts/github-runner-architecture.mdblog-posts/github-runner-architecture.md renamed to docs/archive/blog-posts/github-runner-architecture.md blog-posts/github-runner-architecture.md renamed to docs/archive/blog-posts/github-runner-architecture.md

@@ -0,0 +1,55 @@
++++
+title="Building a GitHub Actions-Style Runner with Firecracker VMs and Knowledge Graph Learning"
+date=2025-12-25
+
+[taxonomies]
+categories = ["Engineering", "Rust", "CI"]
+tags = ["Terraphim", "firecracker", "github-actions", "microvm", "knowledge-graph"]
+[extra]
+toc = true
+comments = true
++++
+
+We built a production-ready GitHub Actions-style workflow runner that combines Firecracker microVM isolation with knowledge-graph-driven learning about what actually works.
+
+<!-- more -->
+
+## Overview
+
+The `terraphim_github_runner` crate provides a complete system for:
+
+1. Processing GitHub webhook events into executable workflows
+2. Spawning and managing Firecracker microVMs for isolated command execution
+3. Tracking command execution patterns in a knowledge graph
+4. Learning from success/failure to improve future workflows
+
+## Architecture
+
+### High-Level Data Flow
+
+```text
+GitHub Webhook -> WorkflowContext -> ParsedWorkflow -> SessionManager
+                                              |
+                                          Create VM
+                                              |
+                              Execute Commands (VmCommandExecutor)
+                                              |
+                            +-----------------+-----------------+
+                            |                                   |
+                    LearningCoordinator                  CommandKnowledgeGraph
+                    (success/failure stats)              (pattern learning)
+```
+
+## Key Components
+
+### 1. VM Executor
+
+The VM executor is the bridge to Firecracker.
+
+### 2. Command Knowledge Graph
+
+The knowledge graph tracks command execution patterns.
+
+## Source Material
+
+- Source: `docs/archive/blog-posts/github-runner-architecture.md`

marketing/blog-post.md docs/archive/marketing/blog-post.mdmarketing/blog-post.md renamed to docs/archive/marketing/blog-post.md

@@ -0,0 +1,43 @@
++++
+title="Building Privacy-First AI Tools with X Integration"
+date=2026-01-19
+
+[taxonomies]
+categories = ["Engineering", "AI", "Product"]
+tags = ["Terraphim", "privacy", "local-first", "knowledge-graph", "x"]
+[extra]
+toc = true
+comments = true
++++
+
+Knowledge work is fragmented across tools. We are building Terraphim to unify search across that mess without forcing you to upload your private data to cloud AI services.
+
+<!-- more -->
+
+## The Knowledge Fragmentation Problem
+
+You remember reading a specific solution. You know it exists. But it is scattered across:
+
+- local notes
+- team docs
+- code repositories
+- public Q&A
+- social technical threads
+
+A local-first assistant is only useful if it can connect those sources in one place.
+
+## The Privacy Dilemma
+
+Most assistants solve search by centralizing your data on external servers.
+
+Terraphim is designed to run locally. Your data stays on your machine.
+
+## Why X Integration
+
+Technical discussions happen in threads long before they become "official docs".
+
+We want to index the specific X threads you care about into your local knowledge graph so they become searchable alongside your private notes.
+
+## Source Material
+
+- Source: `docs/archive/marketing/blog-post.md`