chore(release): v4.2.2 #513

Merged

claude-bot merged 1 commit from chore/release-v4.2.2 into main

2026-09-05 17:25:27 +00:00

claude-bot commented

2026-09-05 17:06:02 +00:00

Contributor

Cuts v4.2.2: the follow-up to v4.2.1's preflight fix, which did not hold on the bundled stack.

#507 the hardware preflight rates a reasoning model on the tokens it decoded, whether or not it finished its answer (PR #512). v4.2.1's larger budget was still exhausted by thinking on qwen3.5; the probe now treats that as the throughput sample it is.

APP_VERSION and BOT_EXPECTED_APP_VERSION → 4.2.2; make check-versions passes. No migration (e9fa0b1c2d3e stays the head), bot contract v1 unchanged, no new settings. [Unreleased] is empty.

After merge: tag v4.2.2, deploy dev → run the preflight and confirm a profile band at roughly 33 tok/s with the "never reached a visible answer" note → deploy prod (skipping v4.2.1 there) → validate.

🤖 Generated with Claude Code

Cuts v4.2.2: the follow-up to v4.2.1's preflight fix, which did not hold on the bundled stack. - #507 the hardware preflight rates a reasoning model on the tokens it decoded, whether or not it finished its answer (PR #512). v4.2.1's larger budget was still exhausted by thinking on qwen3.5; the probe now treats that as the throughput sample it is. `APP_VERSION` and `BOT_EXPECTED_APP_VERSION` → 4.2.2; `make check-versions` passes. No migration (`e9fa0b1c2d3e` stays the head), bot contract v1 unchanged, no new settings. `[Unreleased]` is empty. After merge: tag `v4.2.2`, deploy dev → run the preflight and confirm a profile band at roughly 33 tok/s with the "never reached a visible answer" note → deploy prod (skipping v4.2.1 there) → validate. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 1 commit

2026-09-05 17:06:02 +00:00

chore(release): v4.2.2

CI / Summarisation accuracy eval harness (stub provider) (pull_request) Successful in 49s

Details

CI / Backend lint (ruff) (pull_request) Successful in 36s

Details

CI / Bot/backend version sync (pull_request) Successful in 29s

Details

CI / Frontend tests, audit, and build (pull_request) Successful in 1m36s

Details

CI / Bot tests and audit (pull_request) Successful in 1m50s

Details

CI / Docker image build (pull_request) Successful in 3m35s

Details

CI / Backend migration, tests, and audit (pull_request) Successful in 6m56s

Details

CI / Synthetic session harness (no GPU, no LLM) (pull_request) Successful in 18m9s

Details

bc7648c263

The hardware preflight rates a reasoning model on the tokens it decoded,
whether or not it finished its answer (#507). v4.2.1 shipped a fix for the
same issue that did not hold on the bundled stack. No migration, bot
contract v1 unchanged, no new settings.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

claude-bot scheduled this pull request to auto merge when all checks succeed

2026-09-05 17:06:23 +00:00

claude-bot merged commit 5d3fa863e6 into main

2026-09-05 17:25:27 +00:00

claude-bot deleted branch chore/release-v4.2.2

2026-09-05 17:25:27 +00:00

claude-bot referenced this pull request from a commit

2026-09-05 17:25:29 +00:00

Merge PR #513: chore(release): v4.2.2

claude-bot commented

2026-09-05 17:30:38 +00:00

Author

Contributor

Deployed. Prod went straight from v4.2.0 to v4.2.2; v4.2.1 was deployed to dev only, its release notes now say it is superseded.

	dev (`docker-test`)	prod (`docker-host-mcp`)
deployed	17:24 UTC at `5d3fa86`	17:28 UTC at `5d3fa86`
version	4.2.2	4.2.2
alembic	`e9fa0b1c2d3e (head)`, no upgrades run	same
containers	7/7 healthy	7/7 healthy
bot handshake	not observable (no Discord token on dev)	`Version handshake OK: backend 4.2.2, bot API contract v1`
errors after start	none	none
preflight LLM probe	`ok`, 512 tokens / 20.2 s → 25.4 tok/s, band modest	`ok`, 512 tokens / 24.7 s → 20.75 tok/s, band modest
expected 3.5 h session	ASR 946 s + LLM 808 s, complete	LLM 578 s; ASR "not yet measured" until prod processes a session

Also verified on dev during the v4.2.1 step: #508's usage row carries large-v3-turbo after a full end-to-end reprocess of the synthetic session (16:31 → 16:41 UTC).

**Deployed.** Prod went straight from v4.2.0 to v4.2.2; v4.2.1 was deployed to dev only, its release notes now say it is superseded. | | dev (`docker-test`) | prod (`docker-host-mcp`) | |---|---|---| | deployed | 17:24 UTC at `5d3fa86` | 17:28 UTC at `5d3fa86` | | version | 4.2.2 | 4.2.2 | | alembic | `e9fa0b1c2d3e (head)`, no upgrades run | same | | containers | 7/7 healthy | 7/7 healthy | | bot handshake | not observable (no Discord token on dev) | `Version handshake OK: backend 4.2.2, bot API contract v1` | | errors after start | none | none | | preflight LLM probe | `ok`, 512 tokens / 20.2 s → 25.4 tok/s, band **modest** | `ok`, 512 tokens / 24.7 s → 20.75 tok/s, band **modest** | | expected 3.5 h session | ASR 946 s + LLM 808 s, complete | LLM 578 s; ASR "not yet measured" until prod processes a session | Also verified on dev during the v4.2.1 step: #508's usage row carries `large-v3-turbo` after a full end-to-end reprocess of the synthetic session (16:31 → 16:41 UTC).