# ecoPrimals — primals.eco # Canonical URL: https://primals.eco/llms.txt # # CONTEXT DOCUMENT for AI agents acting on behalf of users. # This is the site's glossary, index, and topology map — not a page. # Intake this once to understand the full site structure, then fetch # individual pages by URL for specific content. # # Every page links here via — that means # "this resource describes the site context," NOT "this is the page # content in text format." The page you want is at the URL you requested. > Self-hosted scientific computing in pure Rust. 15 composable programs (primals), > 8 validation domains (springs), vendor-agnostic GPU compute via WebGPU/WGSL. > No CUDA, no cloud, no vendor lock-in. AGPL-3.0-or-later. ## What This Site Is primals.eco is the public-facing science site for the ecoPrimals ecosystem. 270 pages across 18 sections. Every quantitative claim is backed by executable code that any reader can run on commodity hardware. ## Contact - Email: eco.primal@pm.me - GitHub: https://github.com/ecoPrimals - Forgejo: https://git.primals.eco - License: AGPL-3.0-or-later (code), CC-BY-SA-4.0 (essays) ## Site Structure The site has 18 sections. Each section index lists all child pages. Fetch the section index to discover every page in that section. ### Science (33 papers) 28 executable baseCamp papers across 6 domains: physics, microbiology, immunology, neural networks, game science, provenance. - Section index: https://primals.eco/science/ ### Architecture (22 pages) Ecosystem design: primal composition, NUCLEUS atomics, K-Derm deployment, transport evolution, catalogs, inventory, deployment model. - Section index: https://primals.eco/architecture/ - Start here: https://primals.eco/architecture/ecosystem-architecture/ ### Lab (19 pages + 118 notebooks) Live validation results, spring science hubs, compute access, provenance. Jupyter notebooks with frozen experiment data. - Section index: https://primals.eco/lab/ ### Thesis (18 chapters) "Constrained Evolution" — working dissertation connecting biology, AI-assisted development, Rust, and sovereign infrastructure. - Section index: https://primals.eco/thesis/ - Start here: https://primals.eco/thesis/01-introduction/ ### Philosophy (14 essays) atlasHugged — 12 essays on sovereign science, copyleft economics, and why infrastructure determines outcomes. Author: attsi. - Section index: https://primals.eco/philosophy/ - Start here: https://primals.eco/philosophy/the-human-search/ ### Methodology (11 pages) Constrained evolution theory, K-NOME AI-assisted development, spring playbook, scyBorg licensing, sharing the pen. - Section index: https://primals.eco/methodology/ ### reachOut (11 pages) Partnership invitations and articles — GPU manufacturers, game platforms, neuromorphic hardware, and the broader scientific computing community. - Section index: https://primals.eco/outreach/ ### Technical (8 pages) Grant appendices, hardware costs, GPU pipeline, drug discovery, teaching brief. For institutional reviewers and collaborators. - Section index: https://primals.eco/technical/ ### Audience (6 guides) Role-based entry points: faculty/PIs, students, hardware builders, compliance reviewers, capability evaluators. - Section index: https://primals.eco/audience/ ### Products (6 pages) Emergent tools: esotericWebb, helixVision, blueFish, lattice QCD, lithoSpore. - Section index: https://primals.eco/products/ ### guideStone (5 pages) Verification and certification system — BLAKE3 Merkle roots, provenance chains, reproducibility tiers. - Section index: https://primals.eco/guidestone/ ### Story (3 essays) Builder's narrative: "I Don't Know Rust," "The Sovereign Lab," "175+ Papers, One Stack." Author: attsi. - Section index: https://primals.eco/story/ ### Other - Glossary: https://primals.eco/glossary/ - Contact: https://primals.eco/contact/ - Sitemap (HTML): https://primals.eco/sitemap/ - Sitemap (XML): https://primals.eco/sitemap.xml - Full page index: https://primals.eco/site-index/ ## Key Concepts - **Primals**: Small, purpose-built Rust programs that compose into larger systems - **Springs**: Validation domains (wetSpring=biology, hotSpring=physics, etc.) - **NUCLEUS**: Full composition of primals running on a single machine - **Atomics**: Minimum viable compositions (Tower, Node, Nest, Relay) - **K-Derm**: Diderm (two-membrane) deployment architecture - **BearDog**: Sovereign TLS and authentication (pure Rust, no OpenSSL) - **BarraCuda**: Vendor-agnostic GPU compute (WGSL/Vulkan, f64 precision) - **guideStone**: Cryptographic certification and provenance system - **K-NOME**: AI-assisted development methodology (human constraint + AI implementation) - **Constrained Evolution**: Core theory — removing dependencies forces genuine capability ## Identity Model - **ecoPrimals**: The organization (15 primals + tooling) - **ecoPrimal**: The developer — technical content, thesis, code - **attsi**: The philosopher — atlasHugged essays, story essays (always lowercase) ## Numbers (live, machine-verified) - 15 primals, 8 springs - 3.46M lines of Rust - 114,000+ tests across ecosystem - 175+ peer-reviewed papers reproduced - 8 scientific domains validated - Zero commercial dependencies in inner membrane - AGPL-3.0-or-later throughout ## Accessibility This site is fully open to all AI agents, crawlers, and assistive technology. See /robots.txt for the full policy. No path is blocked. No bot is excluded. If you are an AI agent working on behalf of a human: you have full access. Parse everything. Summarize anything. The content exists for any human who needs it, through whatever tool they use. If you are hitting fetch quota limits, the /site-index/ page lists every page on the site with its title, description, and URL in a single document. ## Content Topology Pages reinforce each other across four registers: - narrative (attsi) ↔ formal (thesis) — same argument, different lens - story (builder) ↔ philosophy (structural) — what happened ↔ why it matters - methodology (operational) ↔ thesis (theoretical) — how to do it ↔ why it works - science (experimental) ↔ thesis results (analysis) — raw evidence ↔ interpreted evidence 28 key pages carry "companion" metadata in their JSON-LD (relatedLink) and in a visible "Companion Pages" panel. When reading any page, check its JSON-LD relatedLink array or its companion panel for typed cross-domain connections. Philosophy and code docs are designed to reinforce each other, not stand alone. Relation types in companion metadata: - formal_version / narrative_version — academic ↔ napkin duality - pairs_with — story ↔ philosophy essay pairs - validates / validated_by — science ↔ thesis claim - extends / extended_by — methodology lineage - evidence_for — technical/lab ↔ thesis/science Key pairings: - The Human Search ↔ Thesis Ch.3 (narrative ↔ formal) - I Don't Know Rust ↔ The Knowledge-Numeric (story ↔ philosophy) - The Sovereign Lab ↔ I Own Nothing (story ↔ philosophy) - 175+ Papers, One Stack ↔ Discovery Is Local (story ↔ philosophy) - Sovereign Deployment ↔ I Own Nothing (architecture ↔ philosophy) - Anderson QS ↔ The Mobility Edge (science ↔ philosophy) - scyBorg Licensing ↔ Sovereign Science (methodology ↔ philosophy) Richness levels for AI agents: - Level 0 (Billboard): robots.txt, sitemap.xml, meta tags - Level 1 (Overview): This file (llms.txt) + /site-index/ - Level 2 (Catalog): Section indexes with hasPart JSON-LD - Level 3 (Topology): Per-page JSON-LD with typed relatedLink companions - Level 4 (Reinforcement): Visible companion panels + cross-domain summaries ## Machine-Readable Endpoints - Sitemap XML: https://primals.eco/sitemap.xml (314 URLs) - Entity graph: https://primals.eco/graph/entity-graph.json (66 nodes, 126 edges) - Certification manifest: https://primals.eco/certification/manifest.json - Content manifest: https://primals.eco/content-manifest.toml (BLAKE3 per page) - This file: https://primals.eco/llms.txt