Orkestia Reference

Appearance

Proud to collaborate with Microsoft for Startups
Proud to collaborate with Microsoft for Startups

dgi.reasoning-process-synapse-step ​

Process synapse - prepare, call LLM, return output

Process synapse step for reasoning DAG workflow.

Uses services to:

  1. Get synapse from state
  2. Build memory from state
  3. Call LLM via LLMService
  4. Return output and interaction log

Overview ​

PropertyValue
Workflow typeLinear
LibraryApp-dgi
Version1.0

Input Schema ​

FieldTypeRequiredDefaultDescription
skipbooleanNo—If True, skip processing this synapse
synapse_uuidstringNo—Synapse UUID to process (required if not skipping)
synapsesjsonNo—Synapses dict (from state)
memoryjsonNo—Memory state
organization_uuidstringNo—Public org UUID; the internal int is resolved transiently for LLM config
workflow_run_idstringNo—Engine-stamped DAG step run id

Output Schema ​

FieldTypeRequiredDefaultDescription
messagestringNo—User message to reason about
contextjsonNo—Reasoning context (history, organization_uuid, ...)
max_depthintegerNo—Maximum reasoning depth
workflow_run_uuidstringNo—Engine-injected DAG run id
workflow_run_idstringNo—Engine-stamped DAG step run id (E1)
organization_uuidstringNo—Org UUID (public surface)
parent_state_datajsonNo—Parent workflow state_data
reasoningjsonNo——
queuejsonNo——
synapsesjsonNo——
memoryjsonNo——
confidencejsonNo——
conversation_lockjsonNo——
skillsjsonNo——
branchesjsonNo——
virtual_workflowjsonNo——
compiled_configjsonNo——
initjsonNo——
handle_outputjsonNo——
finalizejsonNo——
llm_interactionjsonNo——
synapse_uuidstringNo——
outputjsonNo——
skipbooleanNo——
reasonstringNo——
final_answerstringNo—Final reasoning answer
skills_appliedlistNo——
pending_skillslistNo——
errorstringNo——
error_typestringNo——
failure_reasonstringNo——
failure_typestringNo——
failed_actionstringNo——
compensation_triggerstringNo——
failed_stepstringNo——
failed_at_statestringNo——
failed_layerintegerNo——
comp_queuejsonNo——
comp_current_layerjsonNo——
comp_retry_countintegerNo——
process_synapse_1jsonNo——
handle_output_1jsonNo——
process_synapse_2jsonNo——
handle_output_2jsonNo——
process_synapse_3jsonNo——
handle_output_3jsonNo——
process_synapse_4jsonNo——
handle_output_4jsonNo——
process_synapse_5jsonNo——
handle_output_5jsonNo——
process_synapse_6jsonNo——
handle_output_6jsonNo——
process_synapse_7jsonNo——
handle_output_7jsonNo——
process_synapse_8jsonNo——
handle_output_8jsonNo——
process_synapse_9jsonNo——
handle_output_9jsonNo——
process_synapse_10jsonNo——
handle_output_10jsonNo——
process_synapse_11jsonNo——
handle_output_11jsonNo——
process_synapse_12jsonNo——
handle_output_12jsonNo——

States ​

StateInitialTerminalSuccessAuto-advanceDescription
pendingYesNo—run—
runNoNo—complete—
completedNoYesYes——
failedNoYesNo——

State Diagram ​

Transitions ​

FromActionToDescription
pendingrunrun—
runcompletecompleted—
* (any state)failfailed—

API Usage ​

bash
POST /api/workflows/start
Content-Type: application/json

{
  "workflow_type": "dgi.reasoning-process-synapse-step",
  "initial_data": {
    "skip": true,
    "synapse_uuid": "value",
    "synapses": "value",
    "memory": "value"
  }
}