Skip to content

Human Decision Protocol

Objective

Turn ATLANTYQA philosophy into mandatory operational behavior: no critical decision is executed without explicit human traceability.

When it applies

  1. Daily operations: runbook, infrastructure, governance-catalog, and governance-doc PRs.
  2. Regulated processes: legal/compliance changes and any normatively sensitive step.

Mandatory sequence

  1. declare decision type: daily operations or regulated process.
  2. declare accountable human approver by role.
  3. record final decision: approved, changes requested, blocked, or waiting for evidence.
  4. declare residual risk and next action.
  5. link technical/operational evidence.
  6. link traceability issue/PR.

PR compact format (user-friendly)

In .github/pull_request_template.md, the protocol is captured in 5 blocks:

  1. decision type (choose 1),
  2. accountable approver (choose 1),
  3. final decision (choose 1),
  4. minimum evidence (3 mandatory checks),
  5. extra legal/compliance check only when type is regulated process.

Clarity rule: blocks 1-3 must have exactly one selected option.

Mobile mode (GitHub App)

For mobile operation, the PR body accepts a compact HDP-* block:

HDP-TYPE: operativa-diaria|proceso-regulado
HDP-APPROVER: loky|angel|bosco|hilt
HDP-DECISION: aprobado|cambios-requeridos|bloqueado|en-espera-evidencia
HDP-TRACE: #123
HDP-EVIDENCE: <evidence-url>
HDP-RISK: bajo|medio|alto
HDP-NEXT: <next-action>
HDP-COMPLIANCE: yes

HDP-COMPLIANCE: yes is required only for proceso-regulado.

Corporate role matrix

Role Minimum responsibility Compliance signal
L0KY (Lead) Prioritize and close governance decision PR with explicit final decision
Angel (Junior) Execute runbook and attach evidence Linked reproducible output
Bosco (Infra IT) Validate daily operations and infrastructure Clear operational status and risks
Consultoria Agent HILT Validate regulated legal step Compliance evidence and approval

Technical enforcement (gates)

  • Contract: .github/contracts/human-decision-protocol-gate.json
  • Workflow: .github/workflows/human-decision-protocol-gate.yml
  • Verifier: scripts/verify/human_decision_protocol_guard.py
  • Mandatory PR template: .github/pull_request_template.md

Expected outcome

  • lower decision ambiguity,
  • faster role-behavior adoption,
  • auditable regulated and daily operations.