Agent Skills: Glass Line

Physical substrate layer for Plurigrid ASI — co-deployed fiber optic + geothermal infrastructure providing sensing, communication, energy, and materials extraction through a single bore.

UncategorizedID: plurigrid/asi/glass-line

Install this agent skill to your local

pnpm dlx add-skill https://github.com/plurigrid/asi/tree/HEAD/plugins/asi/skills/glass-line

Skill Files

Browse the full folder contents for glass-line.

Download Skill

Loading file tree…

plugins/asi/skills/glass-line/SKILL.md

Skill Metadata

Name
glass-line
Description
Physical substrate layer for Plurigrid ASI — co-deployed fiber optic + geothermal infrastructure providing sensing, communication, energy, and materials extraction through a single bore.

Glass Line

The bore hole is the conduit. The fiber is the sensor. The heat is the energy. The observation channel and the communication channel are the same physical object.

Position in ASI Lattice

                    ┌─────────────────┐
                    │  glass-bead-game │
                    │  (synthesis)     │
                    └────────┬────────┘
                             │
         ┌───────────────────┼───────────────────┐
         │                   │                   │
┌────────▼────────┐ ┌────────▼────────┐ ┌────────▼────────┐
│  world-hopping  │ │  bisimulation   │ │  triad-interleave│
│  (navigation)   │ │  (dispersal)    │ │  (scheduling)    │
└────────┬────────┘ └────────┬────────┘ └────────┬────────┘
         │                   │                   │
         └───────────────────┼───────────────────┘
                             │
                    ┌────────▼────────┐
                    │     gay-mcp      │
                    │  (coloring)      │
                    └────────┬────────┘
                             │
                    ┌────────▼────────┐
                    │     acsets       │
                    │  (data model)    │
                    └────────┬────────┘
                             │
                    ┌────────▼────────┐
                    │   glass-line     │  ← NEW: physical substrate
                    │  (substrate)     │
                    └─────────────────┘

Four Outputs From One Bore

| Output | Mechanism | Maps To | |--------|-----------|---------| | Energy | Geothermal heat → turbine or heat exchange | DGX Spark cluster power + cooling | | Communication | Glass fiber in bore casing | Low-latency backhaul for hamming swarm | | Sensing | Distributed Temperature Sensing (DTS) via Rayleigh/Brillouin scattering | Real-time thermal gradient monitoring | | Materials | Brine extraction (lithium, rare earths) | Hardware supply chain (Cornwall model) |

Schema

@present SchGlassLine(FreeSchema) begin
  Bore::Ob
  Fiber::Ob
  Sensor::Ob
  Well::Ob

  # A bore contains fibers and connects to wells
  contains::Hom(Bore, Fiber)
  taps::Hom(Bore, Well)

  # A fiber is simultaneously a sensor and a communication channel
  senses::Hom(Fiber, Sensor)
  communicates::Hom(Fiber, Fiber)  # self-referential: the medium IS the message

  # Attributes
  Depth::AttrType
  Temperature::AttrType
  Wavelength::AttrType
  Trit::AttrType

  depth::Attr(Bore, Depth)
  thermal_gradient::Attr(Well, Temperature)
  wavelength::Attr(Fiber, Wavelength)     # 1550nm C-band for telecom, 1064nm for DTS
  trit::Attr(Bore, Trit)                  # GF(3) coloring of physical sites
end

DTS (Distributed Temperature Sensing)

The fiber IS the sensor. No separate instruments needed.

Technique          Resolution    Range     Mechanism
─────────────────────────────────────────────────────
Raman DTS          1m spatial    10km      Anti-Stokes/Stokes ratio
Brillouin OTDR     1m spatial    50km      Frequency shift ∝ temperature
Rayleigh OFDR      1mm spatial   70m       Phase-sensitive backscatter

For a geothermal bore (typically 2-5km depth):

  • Raman DTS at 1550nm: continuous thermal profile of entire bore
  • Same fiber carries 100Gbps+ telecom in separate wavelength band (WDM)
  • Temperature data streams to DuckDB via MQTT → glass-line sensor table

Site Selection Criteria

-- Optimal co-location: geothermal gradient + fiber trunk + compute
SELECT site, geothermal_gradient_c_per_km,
       distance_to_fiber_trunk_km,
       distance_to_compute_facility_km,
       (geothermal_gradient_c_per_km * 10
        - distance_to_fiber_trunk_km
        - distance_to_compute_facility_km * 2) AS score
FROM candidate_sites
WHERE geothermal_gradient_c_per_km > 30  -- minimum viable gradient
  AND distance_to_fiber_trunk_km < 50
ORDER BY score DESC;

Known Candidate Regions

| Region | Gradient | Fiber | Compute | Notes | |--------|----------|-------|---------|-------| | Portland/Cascadia | 40-60°C/km | Major hub (NWAX) | Existing warehouse | Your DGX cluster | | Cornwall UK | 35-40°C/km | Subsea cables | New facility | Lithium co-extraction proven | | Reykjavik | 100+°C/km | IRIS submarine | Verne Global | Already operational | | Nevada/Great Basin | 50-80°C/km | Las Vegas trunk | Switch datacenters | BLM land available | | Pennsylvania mines | Variable | Northeast corridor | Planned 13GW DCs | Abandoned mine cooling |

Integration With Hamming Swarm

Each physical glass-line site becomes a world in the 26-letter mesh:

# A glass-line site binds to a world wallet
class GlassLineSite:
    def __init__(self, letter: str, bore_depth_m: float, fiber_count: int):
        self.letter = letter
        self.world_wallet = WORLD_WALLETS[letter]
        self.bore_depth = bore_depth_m
        self.fiber_count = fiber_count
        self.dts_stream = None  # MQTT topic for thermal data

    def bind_to_swarm(self, mesh: HammingSwarm):
        """Physical site joins the multisig mesh"""
        # The site's thermal output backs the world's DeFi position
        # Geothermal energy production → staking yield analogy:
        # constant baseload output, no intermittency,
        # 90%+ capacity factor (like Amnis stAPT stability)
        mesh.register_site(self.letter, self)

    def sense(self) -> dict:
        """DTS reading from fiber"""
        # Returns temperature profile along entire bore
        # This IS the observation — no separate measurement needed
        return self.dts_stream.latest()

Energy Economics

Geothermal LCOE:    $0.04-0.08/kWh (baseload, 90%+ capacity factor)
Grid power (US avg): $0.12/kWh
DGX Spark (3 nodes): ~6kW sustained
Annual energy cost:
  Grid:        6kW × 8760h × $0.12 = $6,307/yr
  Geothermal:  6kW × 8760h × $0.05 = $2,628/yr
  Savings:     $3,679/yr (~58%)

Cooling savings (ground loop vs HVAC):
  ~40% reduction in cooling energy
  Additional $1,500-2,500/yr savings

Total annual savings: ~$5,000-6,000/yr
  = ~700 APT/yr at current price
  > 5x the entire DeFi yield (10.8 APT/yr)

The physical substrate dominates the digital yield.

GF(3) Triad

| Trit | Layer | Role | |------|-------|------| | -1 | glass-line | Physical substrate (sensing, energy, materials) | | 0 | acsets + gay-mcp | Data model + coloring (digital structure) | | +1 | glass-bead-game | Synthesis (emergent coordination) |

Conservation: the physical (-1) grounds the digital (0) which enables the emergent (+1).

Monitoring

# Stream DTS data to DuckDB
mosquitto_sub -t "glassline/+/dts" | \
  duckdb ~/i.duckdb -c "
    INSERT INTO glass_line_dts
    SELECT * FROM read_json('/dev/stdin', auto_detect=true);"

# Thermal gradient alert
duckdb ~/i.duckdb -c "
  SELECT site, depth_m, temp_c,
         temp_c - LAG(temp_c) OVER (ORDER BY depth_m) AS gradient
  FROM glass_line_dts
  WHERE site = 'portland'
  ORDER BY depth_m DESC LIMIT 20;"

# Energy production vs DeFi yield comparison
duckdb ~/i.duckdb -c "
  SELECT
    'geothermal' AS source, annual_kwh * 0.05 AS annual_usd,
    annual_kwh * 0.05 / 7.5 AS annual_apt  -- at $7.50/APT
  FROM glass_line_sites
  UNION ALL
  SELECT 'defi_yield', 10.84 * 7.5, 10.84
  FROM (SELECT 1);"

Related Skills

  • plurigrid-asi-integrated — parent lattice
  • glass-bead-game — synthesis layer above
  • acsets — data model for bore/fiber/sensor schema
  • defillama-api — DeFi yield comparison
  • duckdb-ies — simultaneity_surfaces view for co-temporal sensing
  • warehouse-network — DGX cluster coordination
  • gx10-offload — compute offload to cluster nodes