TypeDB Notebook Skill

Use this skill to store and retrieve knowledge in the Alhazen TypeDB knowledge graph. This allows you to remember information about papers, create notes, and recall them later.

When to use: "remember this", "save this", "note that", "store", "don't forget", "what do I know about", "recall", "find notes about", "retrieve", "create collection", "build corpus", "gather papers", "classify", "tag", "synthesize", "compare"

Generic Graph CRUD

This skill also provides a schema-validated, generic graph CRUD engine so agents mutate the knowledge graph without authoring raw TypeQL. Describe writes as declarative, schema-checked ops — typed entities, attributes, and relations — and the engine compiles and runs the TypeQL in one atomic transaction. Commands: describe-schema, create-entity, set-attr, delete-attr, link, unlink, delete-entity, query (guarded read-only), apply (atomic op-list), apply-recipe (named op-list template). See USAGE.md → Generic Graph CRUD Engine for the op vocabulary, bindings, and recipe format.

When to use: "create entity", "add a node", "set attribute", "update field", "link", "connect", "relate", "unlink", "delete entity", "run recipe", "apply ops", "describe the schema for", "what attributes does X have"

Prerequisites

• TypeDB must be running: make db-start
• Dependencies installed: uv sync --all-extras (from project root)

Environment Variables

• TYPEDB_HOST: TypeDB server (default: localhost)
• TYPEDB_PORT: TypeDB port (default: 1729)
• TYPEDB_DATABASE: Database name (default: alhazen_notebook)

Quick Start

# Remember something
uv run python .claude/skills/typedb-notebook/typedb_notebook.py insert-note \
    --subject "paper-xyz789" \
    --content "Key finding: 95% editing efficiency in liver cells."

# Recall it later
uv run python .claude/skills/typedb-notebook/typedb_notebook.py query-notes --subject "paper-xyz789"

Command Output Pattern

uv run emits a VIRTUAL_ENV warning to stderr. Always use 2>/dev/null when piping output to a JSON parser — never 2>&1, which merges the warning into stdout and breaks JSON parsing.

Before executing any commands, read USAGE.md in this directory for the complete command reference, workflows, and data model.