Skip to content
AIGP

RFC-020: Governed Autonomy, Symbolic Intent, and D-DNA Evidence — 19. Agentic AI Profile

AIGP Specification › RFC-020: Governed Autonomy, Symbolic Intent, and D-DNA Evidence › 19. Agentic AI Profile

← 18. Force-Capable Systems · Section index · 20. Governance Failure Modes →

19. Agentic AI Profile

Agentic AI is the precursor to embodied autonomy.

Agentic AI follows a loop:

Goal → Plan → Tool Selection → Action → Observation → Re-plan

Embodied autonomy follows a similar loop:

Mission → Perception → Navigation → Action → Consequence → Re-plan

Therefore, AIGP applies the same governance chain to agentic AI.

Agentic D-DNA must capture:

agentic_ddna_required_fields:
  - agent_id
  - task_id
  - delegated_authority
  - tool_requested
  - tool_scope
  - delegation_token_hash
  - prompt_context_hash
  - memory_context_hash
  - retrieved_context_hashes
  - governance_intent_hash
  - symbolic_constraint_hash
  - decision_result
  - input_hash
  - output_hash
  - tool_result_hash
  - cost_and_token_record

Example:

◎agent.finance → call ◇payment_api
✓ if ⌘role:finance_agent ∧ ⛶scope:approved_vendor
⊘ if amount > limit
#D-DNA{event:tool_call.permitted, hash:sha256:c19e...}

← 18. Force-Capable Systems · Section index · 20. Governance Failure Modes →