Skip to main content

DDL — Dreamineering Domain Language

What language lets humans and agents point at the same thing without translation loss?

DDL is the human-readable wrapper for the agent language. It names the concepts agents use when they plan, act, record, and improve. Use AISL in protocol and schema contexts. Use DDL on readable agent surfaces.

In the OKF meaning graph, DDL names the concept node. A term earns graph use when it has one canonical meaning and, when possible, a home route that resolves to a playbook page.

Wisdom Transfer Job​

The target is not a prettier glossary. The target is wisdom transfer.

A solo operator should not have to remember where the business answer lives. The system should know the term, route, proof state, authority boundary, and next action. DDL gives the human-readable name that lets memory, processing, and operation point at the same thing.

Use this stack:

capture -> DDL name -> OKF route -> symbol/code -> platform or CLI action -> receipt

If the name cannot travel that path, it is not ready to become canonical.

Core Terms​

Agent:

  • an accountable actor that reads intent, uses instruments, acts inside authority, and emits receipts.
  • avoid: treating every software tool as an agent.

Instrument:

  • a thing that measures, gates, records, selects, transmits, or warns.
  • avoid: using instrument as a generic name for any page or script.

Receipt:

  • a proof trail that records what happened, under which scope, with which result.
  • avoid: treating logs as receipts when they do not prove authority or outcome.

Loop:

  • the feedback path from intent through action, proof, learning, and the next intent.
  • avoid: using loop when a one-way task has no learning return.

Capability:

  • a declared job an agent can perform under known input, output, authority, and proof rules.
  • avoid: calling a desire a capability before proof exists.

Delegation:

  • the decision-right layer that says act alone, ask first, escalate, or refuse.
  • avoid: letting permission hide inside tool access.

Decoder:

  • the readable expansion that lets a human or agent turn a compressed symbol, code, or route back into the concept, authority, and action it represents.
  • avoid: treating shorthand as language when only the author knows what it means.

Wisdom Transfer:

  • the act of moving a situated decision pattern from one person, page, agent, or tool into another without losing context, timing, authority, or proof.
  • avoid: calling stored knowledge wisdom before it improves a future decision.

Density​

DDL sits between narrative prose and wire-format symbols:

  • prose explains the idea.
  • DDL names the idea.
  • DML compresses the idea.
  • diagrams place the idea in a system.

The same concept should survive each density change.

Checks​

  • A term has one canonical meaning in this context.
  • An overloaded term names its domain.
  • Avoid aliases are visible when drift is likely.
  • A symbol or code links back to a readable term.
  • New capability names use the same language as the source contract.
  • A term that triggers action names the authority boundary and receipt expectation.

Failure Modes​

  • Dialect split — two pages use different names for the same agent concept.
  • Overloaded term — one word means different things without a qualifier.
  • Brand drift — DDL and AISL get treated as competing standards.
  • Decoder gap — symbols exist without a readable term.
  • Memory-only term — a name helps the author remember, but cannot guide another agent.

Context​

Questions​

Which word would cause the next agent to choose the wrong action?

  • Is the term canonical?
  • Is the domain clear?
  • Is the avoid list clear?
  • Is the compressed code linked?
  • Is the name stable enough to build on?