Tags: 0bserver07/staxtrace

test: the three rate-card assertions that named a model rather than a…

… property

Adding Claude Opus 5 to the rate card broke three tests, each for the same
reason: they pinned a model NAME where the property under test was something
else.

- rate_card_membership_is_identity_not_priceability used `claude-opus-5` as
  its example of 'priceable but not carded' — exactly the state the fix
  removed. It now uses an id the card does not list, and additionally asserts
  that opus-5 is now BOTH, so the test documents the change that broke it.
- The two manifest-order assertions checked that the first key is
  `claude-fable-5`. The property is ORDER, not that model; the first key
  tracks whatever heads [canonical_ids].anthropic, which opus-5 now does.

Back to the 19-failure baseline (the pre-existing stax-core/query set).
clippy 0, fmt 0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Release v0.9.2

- Add /live SPA route so the Live tab survives a direct load/refresh
- Recognise deepseek as tool-capable in the meta-agent sidebar
- Fix legacy-project last-active timestamp skewed by shared history.jsonl
- Refresh README + docs site with current dashboard screenshots
- Bump version to 0.9.2, update CHANGELOG

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

v0.9.1 — five audit fixes

v0.9.0 — Wave 1 of world-class roadmap: 6 specs (file-risk, burn-v2, …

…mode/skill recommenders, live tab, schema spec)

v0.8.0 — meta-agent sidebar + CLI cost-report fix + --ingest flag

v0.7.4 — CI cleanup + drop beta on yield/qa/tags + api docs for playb…

…ack fs

v0.7.3 — Playback v2 (file-browser-at-time-T) + opt-in semantic searc…

…h + Windows CI + drop beta flag

v0.7.2 — pricing coverage round 2 + outcome confidence + optimize par…

…ity lock-in

v0.7.1 — init --install-skills + beta pricing coverage

v0.7.0 — ETL pipeline + handoff