← Skillful-Alhazen

Home

Skillful Alhazen

A TypeDB-powered agentic knowledge notebook — run interactively with Claude Code or deployed persistently via OpenClaw

Prototype software — APIs, schemas, and skill interfaces are subject to change without notice.

“The duty of the man who investigates the writings of scientists, if learning the truth is his goal, is to make himself an enemy of all that he reads, and, applying his mind to the core and margins of its content, attack it from every side.”

— Ibn al-Haytham (Alhazen), 965–1039 AD

What is Skillful Alhazen?

Skillful Alhazen is an agentic curation system with knowledge graph memory. The agent reads scientific papers, disease databases, job postings, news articles, and domain-specific knowledge bases — building structured understanding in a TypeDB knowledge graph that persists across sessions and grows with use.

You never write queries or call APIs directly. The agent handles all of it.

Three layers work together:

  • Agent — Claude Code (interactive) or OpenClaw (persistent service). You talk; the agent curates.
  • TypeDB — ontological memory. The schema defines the concepts the agent reasons about; the data is what it has learned so far.
  • Skills — domain modules combining a TypeDB schema namespace, Python CLI scripts, and agent instructions. Each skill extends what the agent can do and remember in a specific domain.

Core vs. External Skills

Core skills are built into this repository (skills/ directory) and always available:

Skill What it does  
typedb-notebook Remember, recall, and organize knowledge — the base layer of the knowledge graph  
web-search Web search via SearXNG (self-hosted metasearch, no API key needed)  
curation-skill-builder Design and build new TypeDB-backed curation skills; 6-phase system design framework  
tech-recon Systematic technology investigation against user-defined success criteria ✓

External skills are cloned from sciknow-io/alhazen-skill-examples and other repositories on make build-skills:

Skill Domain Dashboard
scientific-literature Multi-source literature search + semantic search (PubMed, Europe PMC, OpenAlex, bioRxiv) —
alg-precision-therapeutics Rare disease mechanism of harm and therapeutic landscape from a MONDO diagnosis ✓
literature-trends Trace how explanatory hypotheses evolve across a tagged literature corpus —
they-said-whaaa Track consistency of public figures — transcripts, articles, claim detection, contradictions ✓
jobhunt Job application tracking — ingest postings, fit analysis, skill gap identification ✓
dismech Browse the DisMech knowledge graph (750+ curated disease mechanism entries) —

Three Ways to Run

Mode Setup Best for
[A] Claude Code make build && claude Exploration, skill development, one-off research
[B] OpenClaw on Mac Mini ./deploy/deploy.sh --target-type macmini Persistent local service, Telegram triage, cron foraging
[C] OpenClaw on VPS ./deploy/deploy.sh --target-type vps Always-on production, hardened Linux server

See Deployment for the full A → B → C progression.

Getting started

Architecture

Core skills

Domain skills

Reference