Every ticket as the root of its provenance graph — goal, patch, approval, runs and (as the ticket-class build lands) mandate, manifestations, guarantees, journey, artifacts and children. Rendered through the <<<TicketGraphList>>> tag.
The whole design depends on agents emitting STRUCTURED outputs (dispatch plans, verdicts, work-done envelopes) and TOOL CALLS reliably — local models are weaker at this than Claude. Empirically measure, via Ollama /api/chat tools: how often gemma3 and qwen2.5-coder emit valid tool calls / parseable JSON on our exact schemas across N trials, with vs without few-shot examples and format=json / grammar constraints. This GATES the design: if reliability is too low, we need constrained decoding (GBNF grammars) or a validate-and-retry wrapper. OUTPUT: reliability numbers per model/schema + the required mitigation to hit a usable threshold. Spawns a build ticket.
patch
none
approval
unapproved
runs
no runs recorded
mandate (clauses)
not yet recorded — lands with the ticket-class build
manifestations
not yet recorded — lands with the ticket-class build
guarantees
not yet recorded — lands with the ticket-class build
journey (blunders & successes)
not yet recorded — lands with the ticket-class build
artifacts (forge)
not yet recorded — lands with the ticket-class build