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.
RFC: adopt GROK (xAI) as primary LLM provider via litellm — move off Anthropic
goal
Move the agent stack's model provider from Anthropic to GROK (xAI). Grok's API is OpenAI-compatible, so it slots into the litellm_process proxy (#184) as a backend with ZERO agent-code change — only the model_list yaml changes. INVESTIGATE: (1) grok model ids + capabilities for each ROLE (dispatch/judge, worker/coder, adversary) — which grok tier per role, context windows, tool-calling reliability (the #175 gate applies to grok too); (2) the role->backend map update: fallback and judge move claude->grok, local gemma3/qwen3-coder stay for offline/cost; (3) auth via wallet (grok key), NO telemetry forwarding; (4) rate/cost model vs the abandoned anthropic subscription. DELIVERABLE: the litellm model_list wiring grok per role + a proof call reaching grok through the proxy. Reuse #184; do not build a bespoke grok client.
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