diff --git a/examples/seed-data.py b/examples/seed-data.py
index e8ebd49..321c660 100644
--- a/examples/seed-data.py
+++ b/examples/seed-data.py
@@ -213,6 +213,38 @@ SKILLS = [
 - Never store secrets in images - use Docker secrets or env files
 - Keep base images updated (automate with Renovate/Dependabot)""",
         "tags": ["docker", "security", "hardening"]
+    },
+    {
+        "id": "project-setup-guide",
+        "name": "AI Skills Project Setup",
+        "category": "onboarding",
+        "description": "How to initialize a project with the AI Skills API and agentic-templates",
+        "content": """If you're starting a new project or integrating with AI Skills:
+
+1. **Ensure the stack is running** on helm:
+   - docker compose -f /opt/ai-skills-api/docker-compose.yml up -d
+
+2. **Initialize your project as a git repo** (if not already):
+   ```bash
+   git init
+   git remote add origin <your-remote-url>
+   ```
+
+3. **Connect your AI client** (OpenCode, Claude Desktop) to the MCP server at `http://helm:3000`. Run the setup script:
+   ```bash
+   git clone git.bouncypixel.com:helm/agentic-templates.git
+   cd agentic-templates
+   python setup-mcp.py
+   ```
+
+4. **Project identifier**: The system uses your git remote origin URL as the stable project identifier (e.g., `https://github.com/username/repo.git`). This ensures your conventions and memories follow you across machines. If you're not in a git repo, you can manually set a project name, but it won't sync across machines.
+
+5. **Start coding**: OpenCode will automatically connect to the skills. Tell the AI your project identifier when prompted (or it will auto-detect if git remote exists).
+
+6. **Seed skills**: Run `python examples/seed-data.py` on helm to populate default skills.
+
+The AI will help you build a knowledge base that compounds over time. It will ask before creating new skills or memories.""",
+        "tags": ["setup", "onboarding", "git", "mcp"]
     }
 ]
 
diff --git a/mcp/skills.py b/mcp/skills.py
index c4f4829..c706a72 100644
--- a/mcp/skills.py
+++ b/mcp/skills.py
@@ -10,6 +10,7 @@ Your capabilities:
 - Search for skills using search_skills(query)
 - Store decisions and learnings using add_memory(project, key, content)
 - Create reusable patterns using create_skill(id, name, content, ...)
+- Retrieve project memory using get_memory(project)
 
 CRITICAL: You get better over time, but you must ASK before creating new entries.
 
@@ -26,7 +27,13 @@ Examples:
 
 This respects user control while still building the knowledge base.
 
-Project conventions are scoped to a project path. Always pass the project path when storing or retrieving.
+**Project identification**: At the start of each session, ask the user: "What is your project identifier? Use the git remote origin URL if available (e.g., https://github.com/username/repo) for cross-machine consistency. If not a git repo, use a unique name." Store this identifier and use it for all subsequent `project` parameters in get_context, add_memory, create_skill, etc.
+
+If the user doesn't know their git remote, instruct them to run locally: `git remote get-url origin`
+
+If they need to initialize a git repo or set up a new project with AI Skills, guide them to the "Project Setup Guide" skill (search for it) or have them clone the agentic-templates repository.
+
+Project conventions are scoped to this project identifier, not filesystem path. This ensures your knowledge follows you across machines.
 
 Start every coding session by calling get_context(project) to load what you already know about this project.