diff --git a/README.md b/README.md
index 6339b8e..0e966a1 100644
--- a/README.md
+++ b/README.md
@@ -1,14 +1,11 @@
 # Claw Ideas: Problem Solver Projects
 A collection of architecturally challenging project ideas for developers.
 
-## 1. Automated Code-Review-Pipeline (The Gatekeeper)
-An agent that monitors git PRs and provides intelligent code reviews using LLMs. Focus on architectural feedback rather than just linting.
+These projects are designed for developers looking to grow into "Problem Solver / Engineer" roles by focusing on infrastructure, automation, and intelligent tooling.
 
-## 2. Observability-as-a-Service (Engineering Monitoring)
-Aggregate server, container, and application logs with RAG-based analysis to detect and explain anomalous system behavior.
+## Projects
 
-## 3. Serverless Micro-Stack Manager
-An abstraction layer/CLI to manage complex Docker-Compose stacks or deployments via chat or API. Focus on Developer Experience (DX).
-
-## 4. RAG Knowledge Hub
-A "Second Brain" for your infrastructure, indexing your configs, docs, and notes to answer technical questions about your specific setup.
+- **[Automated Code-Review-Pipeline](./code-review.md)**
+- **[Observability-as-a-Service](./observability.md)**
+- **[Serverless Micro-Stack Manager](./micro-stack.md)**
+- **[RAG Knowledge Hub](./rag-hub.md)**
diff --git a/code-review.md b/code-review.md
new file mode 100644
index 0000000..9dd5053
--- /dev/null
+++ b/code-review.md
@@ -0,0 +1,13 @@
+# Automated Code-Review-Pipeline (The Gatekeeper)
+
+## Problem
+Manual code review is tedious and spesso inconsistent. Architectural anti-patterns often slip through PRs.
+
+## Instructions
+1. **Webhook Integration:** Set up a Gitea webhook that triggers a function on your VPS when a PR is opened.
+2. **Analysis Engine:** Use an LLM (e.g., via Gemini API) to analyze the .
+3. **Prompt Engineering:** Define a system prompt for the agent to focus on *architectural integrity* and *complexity metrics* rather than just syntax checking.
+4. **Feedback Loop:** Have the agent post its review directly back to the PR comments.
+
+## Goal
+Automate the "High-Level Review" stage of your development pipeline.
diff --git a/micro-stack.md b/micro-stack.md
new file mode 100644
index 0000000..d3b5190
--- /dev/null
+++ b/micro-stack.md
@@ -0,0 +1,13 @@
+# Serverless Micro-Stack Manager
+
+## Problem
+Managing multiple Docker Compose projects via SSH/CLI is exhausting and error-prone.
+
+## Instructions
+1. **Abstraction Layer:** Build a thin Go or Python API that wraps  commands.
+2. **ChatOps Interface:** Connect this API to your chat agent (via Discord/Telegram) so you can issue commands like "Deploy auth-service" or "Restart proxy".
+3. **Version Control:** Add logic to pull the latest image/code, backup data volumes, and restart the service atomically.
+4. **Health Check:** Ensure the agent verifies the service is actually "up" (HTTP 200) after the restart before confirming.
+
+## Goal
+Build your own platform-as-a-service (PaaS) on your own VPS.
diff --git a/observability.md b/observability.md
new file mode 100644
index 0000000..37e7d21
--- /dev/null
+++ b/observability.md
@@ -0,0 +1,13 @@
+# Observability-as-a-Service (Engineering Monitoring)
+
+## Problem
+Logs are overwhelming. You only notice issues when the system is already down.
+
+## Instructions
+1. **Aggregator:** Set up a lightweight log forwarder (like Vector or Promtail) to push logs to a central database (Postgres or Elasticsearch).
+2. **Analytics:** Implement a background script that polls for error spikes or latency anomalies.
+3. **RAG Context:** Inject your architecture documentation (READMEs, design docs) into the agent's context so it understands *what* service is doing *what*.
+4. **Actionable Alerts:** When an anomaly occurs, use the agent to analyze the logs *in context* and provide a summary + suggested fix.
+
+## Goal
+Move from "Log Monitoring" to "Intelligent System Health".
diff --git a/rag-hub.md b/rag-hub.md
new file mode 100644
index 0000000..6c5a71e
--- /dev/null
+++ b/rag-hub.md
@@ -0,0 +1,13 @@
+# RAG Knowledge Hub
+
+## Problem
+Technical knowledge about personal infrastructure is scattered across config files, chat history, and stale documentation.
+
+## Instructions
+1. **Data Collection:** Crawl your , , and project directories for configuration files (, , ).
+2. **Embedding:** Use an embedding pipeline to vectorize this knowledge.
+3. **Retrieval:** Set up a web interface or terminal bot that allows querying your "Second Brain".
+4. **Sync Mechanism:** Create a task that updates the vector database whenever a config file is modified (using  or ).
+
+## Goal
+Create an infrastructure-aware assistant that knows your setup intimately.