New to Active Inference
Start with the systematic overview, then see the framework applied to behavior, teams, and decentralized knowledge.
Ant researcher or entomologist
From lab ant genetics to colonies as cognitive models — the entomology thread spans two decades.
Cognitive security
Narrative ecosystems, information management, and agentic cogsec tooling for degraded information environments.
Hiring or collaborating
Who I am, how I work with teams, and where the reusable research infrastructure lives.
- About Daniel Ari Friedman
- CV and structured resume
- Generative Research Teams: Active Inference Compositions
- Active Inference & Behavior Engineering for Teams
- GeneralizedNotationNotation (GNN)
- Pools, Rules, and Tools: A Template-Integrated Resource Pool
- Collaborators
- Media hub: talks, podcasts, and recordings
- Presentations & media domain
What this site is
The public research and software index of Daniel Ari Friedman: hand-maintained curated sources backed by a generated data layer, all in one repository.
This site is the front end of docxology. Curated files (the bibliography, the software catalog, this page) are the sources of truth; everything generated — citation exports, per-work pages, search and agent indexes — is rebuilt from them. The full artifact map, including which files are generated and how to rebuild each one, is in GENERATED.md. The main layers:
- Publications — a unified bibliography of 222 works rendered at publications.html, with a page per work under works/.
- Software — 154 curated software projects at software.html, backed by a full GitHub repository inventory.
- Paper folders — 205 folders under
papers/, 201 with full-text extraction and 8,959 extracted images (recorded in data/current-counts.json). - Art — 943 catalogued pen-and-ink drawings at art.html, one static page per drawing under artworks/ (themed hubs in art-collections/), exported as artworks-index.json and mirrored on Flickr.
- Video — 1,127 videos across the personal and Active Inference Institute channels at videos.html, exported as videos.json.
- Evidence — a claim-level ledger at evidence.html and an executable reproducibility ledger at reproducibility.html.
If you are an agent
Machine-readable entry points, in roughly the order the repository itself recommends them.
- llms.txt — the canonical machine-readable map: every page and dataset, the crawl policy, and the site's source-of-truth rules.
- data/agent-index.json — the stable agent route manifest: routes, dataset schemas (schema registry 1.5), counts, and freshness metadata.
- discovery.html — canonical public identifiers, public API endpoints, and repeatable query recipes for cross-checking claims.
- data/works.json — the machine-readable registry of every work (num, citation key, URL, DOI, paper folder).
- search-index.json — the full site search index; search.html is the human-facing version.
- cite-verify.html — citation metadata and the verification rules: treat curated files as sources of truth and public API counts as freshness checks, not replacements.
Recent work
The most recent additions to the unified bibliography, and where site-level changes are recorded.
Recent additions to the bibliography include:
- Agentic Security and Operating Systems — a deep review and prospectus.
- FractiSkills: One Portable Agent Skill per Page
- Skillarum: Conditionally Reproducible Website-to-Agent-Skill Compilation
- Cognitive Integrity Framework: Practical Applications and Deployment Guide (Part 3; Parts 1–2 also in the bibliography)
Site-level changes are summarized at updates.html (built from the repository changelog) and delivered via the RSS feed.