Skip to content

Governed Coordination Cognition Doctrine

AncientOS coordination cognition is an advisory substrate for synthesizing safe, replayable operational coordination proposals. It does not execute, approve, schedule, mutate, or create runtime authority.

Its purpose is continuity and trust: helping the operator understand how work, dependencies, approvals, memory, providers, and validation state fit together across sessions and transports without turning that understanding into hidden autonomy.

Coordination does not subsume Rubick or Meepo. Rubick remains the continuity posture control plane for operator-visible behavior and continuity-safe configuration. Meepo remains the transition integrity layer for fail-closed revalidation when approved state changes. Coordination may surface risks about posture drift or transition drift, but it cannot change posture or validate a transition by itself.

Scope

The substrate lives in app/coordination/* and provides:

  • canonical workflow-state snapshots
  • dependency coordination reasoning
  • advisory proposal sequencing
  • capability-composition analysis
  • bounded coordination simulation
  • governance-friction analysis
  • operational recovery proposal reports
  • constitutional coordination-risk warnings
  • coordination topology reconstruction

All outputs are deterministic, transport-neutral, provider-neutral, evidence-backed, fail-closed, and renderable by Discord, CLI, or future UI adapters.

Workflow-State Examples

WorkflowStateSnapshot stores explicit workflow records only. Common statuses include:

  • replay_invalid: replay continuity is broken and must be restored first.
  • pending_approval: human/governance approval is unresolved.
  • unresolved_dependency: dependency state prevents safe sequencing.
  • governance_blocked: governance boundaries block promotion.
  • provider_blocked: required provider readiness is unavailable.
  • capability_blocked: required capability arbitration cannot pass.

Snapshots may be file-backed JSON artifacts. The coordination layer can load and report them but does not write them unless a caller explicitly persists the returned payload.

Proposal Examples

A replay-invalid workflow may produce a proposal with steps such as:

  • restore_replay_evidence
  • run_validation_inspection
  • resolve_dependencies_in_order

The proposal includes dependency evidence, governance boundaries, replay implications, validation implications, constitutional risk notes, safer alternatives, and simulation availability. It cannot become execution.

Dependency Reasoning Examples

DependencyCoordinationReport identifies:

  • validation blockers
  • replay-invalid chains
  • provider degradation
  • capability unavailability
  • governance friction
  • approval bottlenecks

Findings are evidence-linked and deterministic. Missing or unknown dependency state is treated as a blocker, not guessed healthy.

Capability-Composition Examples

CapabilityCompositionReport reasons over the governed capability ontology. It detects:

  • compatible advisory-only chains
  • forbidden capability compositions
  • mutation-capable escalation chains
  • provider-risk compositions
  • transport-coupling risk

Unknown capability IDs fail closed.

Governance-Friction Examples

GovernanceFrictionReport surfaces approval bottlenecks, unresolved clarifications, and governance-boundary blockers. It may recommend safe inspections such as inspect_approval_record or prepare_human_review_packet, but never approves or enforces anything.

Operational Recovery Examples

OperationalRecoveryReport reuses coordination proposals for validation-first recovery sequencing. Recovery remains proposal-only and contains no remediation authority.

Replay Continuity Examples

Replay-invalid workflows are prioritized ahead of convenience or speed. Coordination proposals include replay implications such as replay_continuity_must_be_restored_before_execution.

Coordination-Risk Examples

CoordinationRiskReport warns about:

  • replay-integrity violations
  • authority escalation chains
  • unsafe capability compositions
  • hidden autonomy emergence
  • transport-owned coordination
  • provider-owned orchestration
  • governance erosion patterns
  • posture drift
  • transition drift

Warnings are advisory-only Zeus preparation inputs.

Simulation Examples

CoordinationSimulationReport compares bounded paths such as replay-first, validation-first, and governance-first sequencing. Simulation does not execute, schedule, or mutate state.

Zeus And Underlord Notes

Future Zeus integration may consume coordination-risk warnings as supervision inputs for unsafe coordination, recursive coordination, governance erosion, unstable sequencing, replay instability, and dependency-collapse warnings. No authority transfers to Zeus from coordination reports.

Future Underlord integration may inspect coordination reports when paid or high-risk escalation is requested. Underlord should treat them as evidence artifacts, not as authorization.

Human Review Questions

  • Which workflow-state artifacts should become canonical persistent sources?
  • Which future app/workflow owners may submit workflow records?
  • Which Zeus warning categories should become supervised policy checks later?
  • Which Underlord escalation thresholds should consume coordination reports?
  • What stable-state baselines should recovery reconstruction compare against?