Skip to content

Planetary Kernel Overview

Definition

The Planetary Kernel is the deterministic execution substrate of Forge Pool.

It defines:

  • execution semantics
  • workload abstraction
  • sharding discipline
  • verification behavior
  • aggregation integrity
  • replay guarantees
  • artifact truth surfaces

The Kernel is not merely an endpoint.

It is the execution contract that governs all distributed workloads executed across the Forge runtime.


Architectural Position

The Planetary Kernel sits inside the runtime core.

In simplified terms:

text
Web Core

Hub

Planetary Kernel runtime behavior

Agents

Verification + Aggregation

Web Core governs public access, tenancy, billing context, and persistence.

The Kernel governs execution truth.

This distinction is essential.


What the Kernel Owns

The Kernel is responsible for:

  • canonical execution envelope semantics
  • deterministic seed handling
  • workload planning discipline
  • shard-level execution structure
  • verification policy application
  • deterministic reduction behavior
  • replay-grade metadata production
  • artifact truth requirements

If a workload is executed through Forge Pool, it is governed by Kernel doctrine whether or not the caller sees that doctrine directly.


What the Kernel Does Not Own

The Kernel does not own:

  • human identity
  • dashboard sessions
  • project lifecycle management
  • public billing presentation
  • domain-specific UI logic
  • adapter-side translation concerns

Those concerns live outside the Kernel.

This separation keeps execution truth narrow, hard, and defensible.


Core Execution Doctrine

Every Kernel workload must support:

  • deterministic execution semantics
  • reproducible shard planning
  • replay capability
  • traceable outputs
  • verifiable reduction behavior

Execution without replay is incomplete.

Execution without traceability is untrustworthy.

Execution without deterministic doctrine is outside the Kernel model.


Canonical Lifecycle

All workloads follow the same invariant runtime pattern:

  1. request admission
  2. Kernel planning
  3. shard construction
  4. deterministic shard execution
  5. optional verification
  6. deterministic aggregation
  7. result materialization
  8. replay and artifact recording

Workload families differ in semantics.

They do not differ in doctrine.


Workload Abstraction

The Kernel does not understand “industries.”

It understands execution classes.

It operates on:

  • primitive families
  • profiles
  • shard plans
  • seeds
  • verification policies
  • reducers
  • artifact rules

This is what allows one system to support many domains without fragmenting the API surface.


Primitive Families and Profiles

The Kernel resolves workloads through:

  • primitive family name
  • family version
  • profile
  • args
  • seed discipline
  • execution policy

Example:

json
{
  "op": {
    "name": "mc",
    "version": 1,
    "profile": "eta.v1"
  }
}

Primitive families define computational class.

Profiles define workload-specific semantics.

That is the stable model.


Relationship to Adapters

Adapters are translation layers above the Kernel.

They may:

  • normalize domain input
  • construct canonical Kernel requests
  • coordinate multi-stage execution
  • interpret results for downstream systems

But adapters do not define compute truth.

The Kernel remains execution authority.


Why the Kernel Matters

Without a hard execution substrate, distributed compute becomes operational theater:

  • output meaning drifts
  • replay breaks
  • billing becomes disputable
  • verification becomes weak
  • trust erodes

The Planetary Kernel exists so Forge Pool can be more than distributed hardware.

It exists so distributed execution remains correct.


Guarantees

The Kernel is designed to preserve:

  • deterministic output under identical contract conditions
  • reproducible planning and reduction
  • replay-grade traceability
  • policy-aware verification
  • auditable artifact generation

Non-Guarantees

The Kernel does not guarantee:

  • fixed wall time
  • identical agent identity across runs
  • identical routing path
  • static infrastructure topology

Execution truth is stable.

Execution path details may vary.


Final Note

The Planetary Kernel is the reason Forge Pool behaves like an execution system rather than a loose compute network.

It is the contract beneath the API, the Hub, the Agents, and the results.