Model catalog refresh

Keeps auto-swe’s own built-in model catalog current. The platform lists each provider’s live model ids, an agent reads the official pricing pages, and the run opens a draft pull request that cites a source for every changed price and calls out any figure it could not confirm. Point it at your fork and put it on a schedule; a run with nothing to change opens nothing.

Template	model-catalog-refresh
Works on	A git repository, cloned into an isolated Docker sandbox
Where a person acts	Nobody inside the run. The pull request it opens is merged by a person, outside it.

The workflow

Drawn from the template spec at build time. Bookkeeping nodes that only record status are left out; amber nodes are where a person acts.

flowchart TD
  n_listModels["List models"]
  n_checkPrevious{"Check previous?"}
  n_donePreviousOpen(["Done<br/><small>done previous open</small>"])
  n_implement["Implement"]
  n_checkChanges{"Check changes?"}
  n_doneNoChange(["Done<br/><small>done no change</small>"])
  n_runTests["Run tests"]
  n_review["Review"]
  n_checkApproval{"Check approval?"}
  n_checkPriceNotes["Check price notes"]
  n_checkReviewLimit{"Check review limit?"}
  n_priceNotesOk{"Price notes ok?"}
  n_terminateReviewFailed(["Failed<br/><small>review failed</small>"])
  n_reviewFix["Review fix"]
  n_checkScope{"Check scope?"}
  n_openPR["Open PR"]
  n_terminateOutOfScope(["Failed<br/><small>out of scope</small>"])
  n_done(["Done"])
  n_listModels --> n_checkPrevious
  n_checkPrevious -->|"yes"| n_donePreviousOpen
  n_checkPrevious -->|"no"| n_implement
  n_implement --> n_checkChanges
  n_checkChanges -->|"yes"| n_doneNoChange
  n_checkChanges -->|"no"| n_runTests
  n_runTests --> n_review
  n_review --> n_checkApproval
  n_checkApproval -->|"yes"| n_checkPriceNotes
  n_checkApproval -->|"no"| n_checkReviewLimit
  n_checkPriceNotes --> n_priceNotesOk
  n_checkReviewLimit -->|"yes"| n_terminateReviewFailed
  n_checkReviewLimit -->|"no"| n_reviewFix
  n_priceNotesOk -->|"yes"| n_checkScope
  n_priceNotesOk -->|"no"| n_checkReviewLimit
  n_reviewFix --> n_review
  n_checkScope -->|"yes"| n_openPR
  n_checkScope -->|"no"| n_terminateOutOfScope
  n_openPR --> n_done

Maturity: built from the flagship steps

Ships as a built-in template and reuses the engineering steps the flagship flow runs. This particular arrangement of them is not separately exercised end to end against a real repository. See product overview §8.

Use it

Every built-in template is seeded into the workflow library and can be cloned, edited on the visual canvas, and versioned — see architecture for the node types. The spec lives in packages/shared/src/workflow/templates/modelCatalogRefresh.ts.