Dreamineering Symbols
How can an agent compress shared meaning without losing the decoder?
Dreamineering Symbols are the readable home for DML and A&ID notation. DML is the wire-format language. A&ID is the diagram grammar. Both compress DDL terms so agents can coordinate with less token cost and less ambiguity.
Symbol Families
Agent symbols:
- digital agent.
- human authority.
- physical actor.
- orchestrator.
- human-in-the-loop boundary.
Instrument symbols:
- alarm.
- controller.
- gate.
- monitor.
- recorder.
- selector.
- transmitter.
- validator.
Loop symbols:
- closed loop.
- cascade.
- open action.
- learning return.
- feedforward path.
Decision symbols:
- pass.
- fail.
- timed condition.
- event condition.
- quality threshold.
- approval condition.
- kill condition.
Operating Rule
A symbol is valid only when the reader can expand it back to a DDL term.
Use symbols when compression helps coordination. Use prose when the reader lacks the decoder.
Checks
- The symbol has one expansion.
- The expansion maps to a DDL term.
- The family is clear.
- The authority level is explicit when action can affect spend, identity, or trust.
- Unknown symbols fail closed.
Failure Modes
- Pretty diagram — the symbol looks useful but does not govern action.
- Undecoded pack — compressed text saves tokens but loses meaning.
- Authority gap — action symbols omit who can approve or override.
- Collision — one mark points at two meanings.
Context
- depends-on DDL Nomenclature — symbols compress the readable vocabulary.
- pairs-with Language Crosswalk — the crosswalk binds terms, codes, and symbols.
- applies-to Agent Operating Model — the operating loop needs compact but auditable signals.
- proved-by Performance — compression is useful only when it improves retrieval, action, or audit quality.
Questions
Which symbol needs a better decoder?
- Can the reader expand the symbol?
- Does the symbol change an action?
- Is authority visible?
- Does the family prevent collision?
- Should this be prose instead?