docs: stale models.json/auth.json paths and inverted Step provider dialect

[AuroraAeon]

What do you want to change?

Two doc bugs in the shipped docs.

1. models.md, providers.md, quickstart.md, custom-provider.md, SECURITY.md and scripts/smoke/README.md say models.json / auth.json / models-store.json live under ~/.stepcode/agent/. The CLI actually resolves them from the config root: apps/cli/src/main.ts passes modelsPath: join(resolveStepConfigRoot(), "models.json"), getStepAuthPath() returns join(resolveStepConfigRoot(), "auth.json"), and model-runtime.ts:180 puts models-store.json beside modelsPath. docs/step-unified-config-and-mcp.md already documents this move, the other pages were not updated with it.

2. step-integration.md says the built-in Step provider uses the anthropic-messages adapter and that startup "restores its anthropic-messages API". Source says the opposite: STEP_MODEL_API = "openai-completions" and normalizeStepModel repairs a stale overlay away from the legacy Anthropic dialect.

Why?

models.md is the documented way to add a custom model. Following it writes to a path the CLI never reads — that is exactly what happened to me: I created ~/.stepcode/agent/models.json first and it was silently ignored. The dialect paragraph describes the repair backwards, which misleads anyone debugging an endpoint problem.

How? (optional)

9 line-level edits across 7 files, docs only, no code changes. Prepared locally; happy to open a PR once approved with lgtm.

[AuroraAeon]

AI-generated: the issue body above was drafted by an AI assistant (OpenCode) from source-code evidence I verified locally — apps/cli/src/main.ts, packages/coding-agent/src/step/auth.ts, packages/providers/src/step-provider/index.ts, packages/coding-agent/src/core/model-runtime.ts. I reviewed the claims before submitting.

[github-actions]

This issue was auto-closed. All issues from new contributors are auto-closed by default.

Maintainers review auto-closed issues daily and reopen worthwhile ones. Issues that do not meet the quality bar in CONTRIBUTING.md will not be reopened or receive a reply.

If a maintainer replies lgtmi on one of your issues, your future issues will stay open. If a maintainer replies lgtm, your future issues and PRs will stay open. The command must be at the start of the reply (optionally after one or more @username mentions) or at the end.

See CONTRIBUTING.md.