Download ZetaNotes

ZetaNotes Manual

Desktop, CLI, and Codex skill workflows.

1. Install

Install the desktop app if you want a visual knowledge base. Install the CLI if you want shell automation or Codex integration.

Desktop install

  1. Download ZetaNotes_1.0.0_x64_en-US.msi or ZetaNotes_1.0.0_x64-setup.exe.
  2. Run the installer.
  3. Open ZetaNotes. On first launch it creates Wiki directories and an index database.

CLI install

  1. Download zetanotes-cli-1.0.0-windows-x64.zip.
  2. Extract it to a stable folder, for example %LOCALAPPDATA%\ZetaNotes\bin.
  3. Add that folder to PATH.
  4. Run zetanotes-cli --version and zetanotes-cli --json doctor.

2. Desktop basics

The desktop app has a page tree, editor, reading view, graph, review queue, AI Organize panel, and Ask Wiki panel.

  • Knowledge directories: daily, concepts, sources, pages, raw.
  • Block anchor: a stable ID such as ^b-aB12cD34.
  • Block reference: a precise reference such as ((b-aB12cD34)).
  • Page link: a Wiki page link such as [[concepts/transformer]].
  • Provenance: metadata such as source_doc, source_locator, model_id, agent_session_id and confidence.

3. Import and ingest

Import copies a source file into the Wiki raw directory and converts supported files to Markdown when needed. Ingest creates a persistent workflow that can turn raw material into source and concept pages.

  1. Click Import File or use AI Organize -> Smart Import.
  2. Select a PDF, DOCX, Markdown or other supported document.
  3. Review the generated workflow plan.
  4. Apply the workflow when the operations look correct.
  5. Open Review to inspect generated blocks and source metadata.

With an OpenAI-compatible LLM configured, ingest can extract better concepts and summaries. Without an LLM, ZetaNotes uses a conservative local fallback and still produces reviewable operations.

4. Ask and maintain

Ask Wiki searches the local index, collects page and block citations, and asks the configured LLM to answer only from those citations. Without an LLM, it returns a cautious fallback answer with citations.

AI Maintenance checks missing anchors, duplicate anchors, dead references, orphan pages, review status and other health issues. It creates a persistent workflow so you can review before applying changes.

5. Learning assistant

The Desktop Learning Assistant turns a Wiki page or topic into study material grounded in your local Wiki. It is different from Ask Wiki: Ask Wiki answers one question, while Learning Assistant organizes a learning session.

Study the current page

  1. Open a Wiki page, such as pages/learning-demo.md.
  2. Expand the Learning Assistant in the left panel.
  3. Optionally enter: Explain this page for a beginner, focus on examples, and include page/block citations.
  4. Click Study current page.

Generate a quiz

Select a page or enter a topic, then click Generate quiz. You can request multiple-choice, true/false and short-answer questions with answers, explanations and Wiki citations.

Generate a learning path

Select a page or enter a goal, then click Generate learning path. The result should order concepts, pages or blocks and explain why each step comes next.

Save and cite the result

Learning results appear in the Learning view. Use Save to Wiki to write a Markdown study page such as pages/study/self-attention-study.md. The saved page keeps task type, scope, result, citations, warnings and returned model/session metadata.

With no LLM configured, the app shows a conservative local fallback or a clear configuration notice. With an OpenAI-compatible LLM configured, the prompt includes the current page and Wiki evidence so the answer can cite real pages and blocks.

The CLI also exposes dedicated learning commands for automation:

zetanotes-cli learn page pages/learning-demo.md --instructions "Explain for a beginner and include citations" --json
zetanotes-cli learn quiz --topic "self attention" --json
zetanotes-cli learn path --topic "Transformer basics" --json
zetanotes-cli learn page pages/learning-demo.md --output pages/study/learning-demo-study.md --json

The commands are read-only unless --output is provided. Desktop is optimized for interactive learning; CLI is optimized for scripts, Codex skills and batch automation.

6. CLI quick reference

zetanotes-cli --json doctor
zetanotes-cli pages list --json
zetanotes-cli pages read pages/example.md
zetanotes-cli pages write pages/example.md --content "# Example"
zetanotes-cli search "self attention" --json
zetanotes-cli ask "What is self attention?" --json
zetanotes-cli backlinks b-aB12cD34 --json
zetanotes-cli graph --scope concepts/ --json
zetanotes-cli review --status unreviewed --json
zetanotes-cli health --json

Workflow commands

zetanotes-cli import C:\Users\me\Desktop\paper.pdf --json
zetanotes-cli ingest C:\Users\me\Desktop\paper.pdf --instructions "Extract reusable concepts" --json
zetanotes-cli workflows list --json
zetanotes-cli workflows status wiki-ingest-... --json
zetanotes-cli workflows apply wiki-ingest-...
zetanotes-cli maintain --instructions "Fix dead refs and missing anchors" --json

7. Codex skill setup

The bundled skill is named zetanotes-wiki. It lets Codex call zetanotes-cli to inspect, search, ask, ingest or maintain your local Wiki.

Install from the release zip

  1. Download zetanotes-codex-skill-1.0.0.zip.
  2. Extract it so the folder becomes %USERPROFILE%\.codex\skills\zetanotes-wiki\SKILL.md.
  3. Make sure zetanotes-cli.exe is on PATH.
  4. Restart Codex if the skill list does not refresh automatically.

Install from source

cd E:\zetaagent\zetanotes-cli
.\scripts\install-codex-skill.ps1

Useful Codex skill requests

  • Ask: "Use my ZetaNotes Wiki to answer: what is self attention?"
  • Search: "Search my ZetaNotes Wiki for transformer and summarize the relevant blocks."
  • Ingest: "Ingest C:\Users\me\Desktop\paper.pdf into ZetaNotes and create a reviewable workflow."
  • Maintain: "Check my ZetaNotes Wiki for dead refs, missing anchors and orphan pages."

8. Example workflows

Study a paper

  1. Import the PDF.
  2. Run Smart Import with: Extract reusable concepts, key claims, and source-backed summaries.
  3. Review the workflow operations.
  4. Apply the workflow.
  5. Ask: What does this paper say about attention?

Weekly maintenance

  1. Run AI Maintenance.
  2. Use instructions: Fix missing anchors, flag weak provenance, and suggest links between related concepts.
  3. Review operations before applying.
  4. Open Graph to inspect newly connected pages and blocks.