Composition Pipeline: Springs → Primals → Products → Foundation

The Pipeline

The 🔧🦎 ecoPrimals ecosystem is a composition pipeline. Science enters at the springs, flows through primal infrastructure, emerges as products, and proves patterns that 🏛️🌱 projectFOUNDATION takes to institutions.

Published science (papers, databases, NCBI, ILDG)
    ↓ Phase 0: Python reproduction (ground truth)
Springs (7 science domains)
    ↓ Phase 1: Rust cross-validation (parity)
    ↓ Phase 2: GPU acceleration (barraCuda + coralReef)
    ↓ Phase 3: Composition dispatch (toadStool + NUCLEUS)
Primals (15 Rust binaries)
    ↓ compose via deploy graphs
Products (helixVision, esotericWebb, blueFish, lattice QCD)
    ↓ prove patterns under real workloads
foundation (institutional adoption)
    ↓ metallic bonding, HPC federation
Institutional science (ICER, university labs, NIH)

Each arrow is independently verifiable. The Rust matches the Python. The GPU matches the CPU. The composition matches standalone. And every result carries provenance: BLAKE3 content hashes, DAG sessions, permanent ledger entries, ed25519-witnessed attribution braids.

Layer 1: Springs — Where Science Enters

Seven springs cover seven scientific domains. Each follows the same phased validation protocol:

The eighth spring — 🧬♨️ primalSpring — validates the composition layer itself: deploy graphs, BTSP encryption, discovery hierarchy, cross-gate bonding. It is the meta-spring that tests the infrastructure every other spring runs on.

Total: 16,000+ automated checks across 175+ published papers.

Layer 2: Primals — Where Capabilities Live

Springs produce validated kernels. Primals provide the infrastructure that runs those kernels in composition:

Primals communicate over JSON-RPC (Unix domain sockets locally, TCP across gates). All 15 implement BTSP Phase 3 with ChaCha20-Poly1305 AEAD. Products consume primals by capability, not by name — the ⚛️🧬 NUCLEUS substrate is invisible.

Layer 3: Products — Where Composition Becomes Useful

Products are chemical reaction products — emergent capabilities from primal composition. Each product consumes primals via JSON-RPC:

Products don’t reference primals by name in their code. They consume capabilities through the Neural API. If 🐟⚡ barraCuda is replaced by a faster GPU compute primal, products don’t change.

Layer 4: Foundation — Where Proof Becomes Adoption

🏛️🌱 projectFOUNDATION takes what projectNUCLEUS proves and makes the case to institutions:

projectNUCLEUS (deploys + validates)
    ↓ 13/13 primals, 235+ checks, provenance pipeline
    ↓ Cloudflare Tunnel baseline, ABG ionic access
    ↓ Three-layer security pen testing
foundation (institutional adoption)
    ↓ metallic bonding → university HPC clusters
    ↓ ICER mapping → MSU compute resources
    ↓ grant appendix → NIH/NSF/USDA funding

The ⛺📄 baseCamp papers are the evidence. projectNUCLEUS is the proof that the evidence runs on real hardware under real load. foundation is the bridge to institutions that have the hardware to run it at scale.

The ABG pattern: ABG members run workloads on the active gate via JupyterHub. Their science validates the infrastructure under real external load. When they point a PI at their results (via primals.eco/lab or the shared workspace), that PI sees provenance-verified, reproducible science running on commodity hardware — the case for institutional compute allocation.

The Feedback Loop

The pipeline is not linear. It loops:

1. Springs validate against published science
2. Primals compose the validated kernels into a substrate
3. Products exercise the substrate under real workloads
4. Gaps flow upstream via wateringHole handoffs
5. Springs evolve to close the gaps
6. primalSpring absorbs the evolution and updates deploy graphs
7. projectNUCLEUS deploys the new patterns on real hardware

Every ABG workload that succeeds proves the composition layer works. Every gap report that flows upstream improves the primals. The more external load the system handles, the more evolved it becomes.

What Makes This Different

Traditional scientific computing stacks are assembled from unrelated parts: SLURM + CUDA + Python + HPC cluster + cloud storage. Each part has its own versioning, its own licensing, its own security model, its own failure modes.

The 🔧🦎 ecoPrimals stack is grown from a single substrate. Primals share a binary standard ( 1️⃣📦 UniBin), a wire protocol (JSON-RPC), a security model (BTSP), and a distribution mechanism ( 🧬📦 plasmidBin). Springs share a validation protocol (Phase 0-3). Products share a composition model (deploy graphs).

The result: a system where the science, the infrastructure, and the security evolve together — and where every computation carries cryptographic provenance from input to output.