chore(pricing): record 0.39.53 publish snapshot

[pitimon]

PR Goal (one sentence)

Record the LiteLLM pricing-seed snapshot regenerated by the 0.39.53 npm publish (prepublishOnly -> node scripts/build-pricing-seed.cjs), matching the pattern of prior publish-snapshot PRs (e.g. #186).

Scope

• CLI (src/)
• Dashboard (dashboard/)
• Docs / CI / config

Checklist

• npm run ci:local passes, or focused test commands are listed below with rationale
• New dashboard/user-facing strings go through dashboard/src/content/copy.csv (N/A — data file only)
• Commits follow conventional style (feat: / fix: / refactor: / docs: / chore: / test: / ci:)
• PR description explains why, not just what
• Version bump files are in lockstep when release-bound (N/A — no version change, data snapshot only)
• npm publish state is recorded when release-bound: 0.39.53 published and verified on the registry (npm view @ipv9/tokentracker-cli version -> 0.39.53; node scripts/verify-registry-package.cjs 0.39.53 -> ok, dashboardAssets=16, cliHelp=ok)

Why

0.39.53 was published via an authorized local recovery publish (CI Trusted Publishing is still blocked on #179 — GitHub-hosted runner gets npm error 404 on the registry PUT). npm publish runs prepublishOnly, which rebuilds the dashboard and refreshes src/lib/pricing/seed-snapshot.json from live upstream LiteLLM data as a side effect. That refreshed seed is what actually shipped in the published tarball (kept_models 2582 -> 2633, generated_at -> 2026-08-22T15:00:43.511Z) and should be recorded in source so a fresh checkout matches the registry artifact, per CLAUDE.md: "Inspect any tracked pricing-seed change before commit."

Verified the fable/mythos fix (#187/#188) still resolves correctly against the refreshed seed:

claude-auto-pilot-fable-v1-canary => true curated:fuzzy
claude-mythos-v1-canary           => true curated:fuzzy

Codex Context (required when requesting @codex review)

• Delta since last Codex review: N/A, first review on this branch
• Intended behavior / invariants: curated overrides continue to take priority over the refreshed LiteLLM seed for pinned models (Fable/Mythos/Opus 4.8/etc); no matcher logic changed, data-only diff.
• Edge cases covered: re-verified the Claude auto-pilot/canary model ids (fable/mythos) bill $0 — fuzzy pricing match misses non-contiguous substrings #187 regression case against the new seed data (see above).
• Tests run (command + result): npm run ci:local -> full pass (1054 product-suite tests + 4 architecture-guardrail tests, dashboard build+test, all validators/checkers).
• Known gaps / out of scope: does not address release: migrate npm publication to Trusted Publishing OIDC #179 (CI OIDC publish) — that remains open and this release again used the documented local-recovery path.

Regression Test Gate

Most likely regression surface

• A stale/incorrect seed snapshot silently shipping wrong prices for newly-added upstream models.

Verification method (choose at least one)

• npm run ci:local full pass on the exact tree that produced this diff, plus a manual pricing-matcher smoke check against the new seed for the models this release fixed.

Uncovered scope

• Issue release: migrate npm publication to Trusted Publishing OIDC #179 (migrate npm publish to Trusted Publishing/OIDC) remains open; this release used the documented local-recovery path again.

[chatgpt-codex-connector]

To use Codex here, create a Codex account and connect to github.