PTENES
TRACK 3

🚀 Real-world and advanced use

With your second brain set up: how to ask questions using the graph, build a Claude OS, decide between code and documents, and keep the vault alive without letting it become a mess.

4
Modules
24
Topics
~2h30
Duration
0%
Your
progress
You question command center Claude OS consult the vault the vault graph notes projects questions contextual answer

↑ O Claude OS as a command center: your question goes to the center, which queries graph, notes, projects e questions in parallel—and the answers converge back in one place.

Trail map

Detailed content

3.1~40 min

💬 Ask Claude Code using the graph

With the vault in the project, make the agent respond using the map instead of scanning everything.

What it is:

Keep the vault inside (or alongside) the project and tell Claude Code where it is.

Why:

It only uses what it can see — if the vault isn't in context, it's as if it doesn't exist.

Concepts:

Vault path · context · project.

What it is:

Use the god nodes from GRAPH_REPORT.md as a starting point for questions.

Why:

Hubs cover more ground—starting with them brings the rest of the map along.

Concepts:

God node · hub · entry point.

What it is:

Ask it to follow the [[links]] from one concept to its neighbors.

Why:

Connected answers, not isolated ones — the agent moves through neighboring nodes instead of stopping at the first one.

Concepts:

Backlink · neighborhood · navigation.

What it is:

Ready-to-use prompt — "use the vault in <folder> to explain X and its related concepts, citing the notes".

Why:

Directs the agent to the graph instead of letting it scan files blindly.

Concepts:

Prompt · cite notes · scope.

What it is:

graphify --mcp exposes tools query_graph, get_node, shortest_path.

Why:

The agent queries the graph programmatically, without relying only on the markdown files.

Concepts:

MCP · query_graph · resource.

What it is:

Go from the concept to the linked source doc and check that it matches.

Why:

Prevents hallucinations and closes the loop: provenance takes you back to the evidence.

Concepts:

Provenance · verification · source doc.

View Full
3.2~45 min

🏛️ Building a Claude OS command center

Turn the vault into a command center: multiple projects, knowledge, and questions in one place.

What it is:

A main vault that brings projects, content, and imported knowledge together in a coherent system.

Why:

The stack shines when it’s part of a whole—not as disconnected graphs.

Concepts:

Command center · main vault · system.

What it is:

Organize the vault into projects/areas and a folder for graph imports.

Why:

Everything in its place, easy to find—and easy for the agent to navigate.

Concepts:

PARA-like · projects · imports.

What it is:

Each new Graphify goes into an import subfolder and links to what’s already there.

Why:

The brain grows in a controlled, curated way instead of becoming a dump.

Concepts:

Subfolder · connection · curation.

What it is:

Create bridge notes that connect the imported knowledge to your real projects.

Why:

Gets the graph out of a vacuum — the video’s thesis: integrated knowledge is more valuable than isolated knowledge.

Concepts:

Bridge note · context · integration.

What it is:

Use the agent to query, summarize, and update the command center vault.

Why:

The center stays alive, not static—it edits alongside you.

Concepts:

Copilot · query · maintain.

What it is:

For code bases or one-off queries, stopping at Graphify may be better.

Why:

Not every case calls for Obsidian — setting up a hub just for one question is overhead.

Concepts:

Trade-off · silo · simplicity.

View Full
3.3~35 min

🧭 Code vs. documents: when to stop at Graphify

Not every graph needs to go into Obsidian. Learn when to stop.

What it is:

Point Graphify to a code repository (AST extraction via tree-sitter).

Why:

Maps functions/modules without an API key—fast and cheap.

Concepts:

Code base · AST · tree-sitter.

What it is:

Point to docs/PDFs (semantic extraction via LLM).

Why:

Captures concepts and relationships in text—not just the literal structure.

Concepts:

Corpus · LLM · semantics.

What it is:

For exploring a specific code base, graph.html + query are enough.

Why:

Moving it to Obsidian would be overhead when you just want a quick answer.

Concepts:

Stop · graph.html · query.

What it is:

When knowledge needs to live alongside your projects and notes.

Why:

It’s the use case for an integrated second brain — not an isolated graph.

Concepts:

Integrate · vault · larger context.

What it is:

Use graphify update/--watch as the code/docs change.

Why:

The graph doesn’t age—it stays true to the source.

Concepts:

Update · watch · sync.

What it is:

Weigh extraction time, graph noise, and the effort of maintaining the vault.

Why:

A conscious decision, not hype—the cost is real.

Concepts:

Cost · noise · maintenance.

View Full
3.4~30 min

🛠️ Maintenance, updates, and troubleshooting

Keep the graph and vault healthy, and know what to do when something goes wrong.

What it is:

graphify update . reprocesses only what changed; re-export the --obsidian regenerates the vault.

Why:

Keep it current without rebuilding everything from scratch.

Concepts:

Update · re-export · incremental.

What it is:

The Obsidian export is regenerated from scratch—manual edits to the generated notes are lost.

Why:

Keeping your notes in separate folders (imports vs. yours) protects your work.

Concepts:

Regeneration · separation · imports.

What it is:

Run uv tool update-shell and reopen the terminal.

Why:

The binary isn't in the PATH—the shell needs to reload.

Concepts:

PATH · shell · uv.

What it is:

Set ANTHROPIC_API_KEY for headless use; inside the /graphify doesn't need to.

Why:

Distinguishing the two paths avoids a key error where no key is needed.

Concepts:

API key · headless · skill.

What it is:

Clean the source, use --mode deep or re-scope.

Why:

Improves entities/relationships—better corpus, better graph.

Concepts:

Noise · --mode deep · scope.

What it is:

Commit graphify-out/ (or the vault) in git; use the course's own export to save your progress.

Why:

Reproducible and safe — you won't lose what you built.

Concepts:

Git · backup · reproducibility.

View Full