Deployment Model: plasmidBin & BYOB

Repository: github.com/ecoPrimals/plasmidBin — Public

The Problem

Primals are self-contained Rust binaries. Springs validate them. Products compose them. But how do binaries get from the primal source tree to the user’s machine without requiring everyone to compile from source?

The Solution: plasmidBin

🧬📦 plasmidBin is the ecosystem’s binary distribution surface. It is analogous to node_modules for primals — a local deployment cache where pre-built binaries are resolved, verified, and composed.

How It Works

Primal Source → cargo build → harvest.sh → plasmidBin (GitHub) → fetch.sh → Local plasmidBin/ → biomeOS deploy

1. Build: A primal is compiled from source (musl-static PIE for x86_64, aarch64 planned).
2. Harvest: harvest.sh validates the binary (static ELF, stripped), computes blake3 checksums, copies into primals//springs/, and updates checksums.toml. Optionally pushes a GitHub Release.
3. Fetch: Consumers run fetch.sh to download from the latest GitHub Release and verify blake3 checksums (via b3sum).
4. Deploy: 🌿🖥️ biomeOS reads deploy graphs (TOML DAGs) and germinates primals from the local plasmidBin/ directory.

Zero Source Coupling

Products and springs never compile primal source. They consume pre-built binaries only. This is the 📦🔧 BYOB (Bring Your Own Biome) model — a product declares which primals it needs via a deploy graph, fetches them from 🧬📦 plasmidBin, and 🌿🖥️ biomeOS handles the rest.

Source Availability

All binaries distributed through 🧬📦 plasmidBin are under AGPL-3.0-or-later. Per AGPL, corresponding source must be obtainable when binaries are distributed:

• Public primals (songBird, nestGate, toadStool, squirrel, biomeOS, 🪸🌊 coralReef, 🐟⚡ barraCuda, petalTongue, sourDough, bingoCube, rhizoCrypt, sweetGrass, loamSpine): source is on GitHub at github.com/ecoPrimals.
• Remaining primals (bearDog, skunkBat): source is available on request. Each metadata.toml includes a [provenance] built_from field that identifies the source tree. Repositories are progressively published as primals mature.

metadata.toml Format

Every primal in 🧬📦 plasmidBin has a metadata.toml describing its identity, provenance, capabilities, and build artifacts:

[primal]
name = "toadstool"
version = "0.6.0"
domain = "compute"
description = "Universal compute orchestration"
license = "AGPL-3.0-or-later"

[provenance]
built_from = "primals/toadStool"
built_at = "2026-03-15T00:00:00Z"

[compatibility]
min_ipc_version = "3.0"
capabilities = ["compute.dispatch", "compute.discover", "ember.route"]

[builds.x86_64]
binary = "toadstool-x86_64"
target = "x86_64-unknown-linux-musl"
checksum_blake3 = "..."
pie_verified = true
static_linked = true

[genomeBin]
tier = "foundation"
unibin_modes = ["server", "cli", "benchmark"]
default_mode = "server"

[genomeBin.server]
default_port = 9100
env_prefix = "TOADSTOOL"

[genomeBin.service]
restart = "always"
after = ["beardog", "songbird"]

What It Captures

BYOB Composition

Products (gen4, 🌱🏡 sporeGarden) declare their primal dependencies in deploy graphs — TOML DAGs that describe which primals to germinate and how they wire together:

# esotericWebb deploy graph (simplified)
[[node]]
primal = "beardog"
required = true

[[node]]
primal = "songbird"
required = true
depends_on = ["beardog"]

[[node]]
primal = "squirrel"
required = false
depends_on = ["beardog", "songbird"]

[[node]]
primal = "petaltongue"
required = false
depends_on = ["squirrel"]

Products use PrimalBridge (JSON-RPC over discovered sockets) to communicate with germinated primals. Graceful degradation is built in: if an optional primal is unavailable, the product continues with reduced capability.

Composition Presets

🧬📦 plasmidBin includes standard composition presets via ports.env:

Offline Capability

After the initial fetch.sh, 🧬📦 plasmidBin is fully offline-capable. Binaries are local, deploy graphs are local, and 🌿🖥️ biomeOS germinates from the local cache. No cloud, no API keys, no network required at runtime.

Current Inventory

🧬📦 plasmidBin currently tracks 18 entries (12 primals + 6 springs with deployment metadata):

Springs in 🧬📦 plasmidBin have deployment metadata (ports, capabilities) for ⚛️🧬 NUCLEUS integration — they are not just validation workspaces but also deployable science services that register with 🌿🖥️ biomeOS.

🧬📦 plasmidBin is the bridge between “we built it” and “you can run it.” Clone the repo, run fetch.sh, and you have a sovereign computing stack.