chore(release): v1.3.0 - #439
Merged
Merged
Conversation
Promotes 1.3.0rc1, whose only purpose was to verify the everalgo transitive pin against a real PyPI resolution — `make package` installs the built wheel with `--no-deps`, so the pipeline's own smoke cannot catch a resolution break. No `### Upgrade` group: the LanceDB table definitions are byte-identical to v1.2.3, so no rebuild is involved and a plain upgrade is all there is. The vector-metric change is described under Changed instead, since it shifts ranking but requires no action. `[Unreleased]` still compared against v1.1.4, three releases behind; moved to v1.3.0.
Kendrick-Song
approved these changes
Sep 7, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Promotes
1.3.0rc1to the stable line. The RC existed to verify one thing that no gate in this repo can: that a real PyPI resolution — notuv.lock— produces a working install.make packageinstalls the built wheel with--no-deps, so the release pipeline's own smoke is blind to a resolution break, which is how 1.2.3 shipped withPOST /addanswering 500.Changes
CHANGELOG.md## [1.3.0]section;[Unreleased]link moved off v1.1.4 (three releases stale) to v1.3.0pyproject.tomlversion = "1.3.0"docs/openapi.jsoninfo.versiontrackseveros.__version__, andmake lintfails on a stale dumpuv.lockNo
### UpgradegroupDeliberate, per the release skill's "omit the group entirely when a plain
pip install --upgradeis all there is". Verified rather than assumed:git diff v1.2.3..main -- src/everos/infra/persistence/lancedb/tables/is empty, so the physical LanceDB schema is unchanged and no rebuild is involved. The vector-metric change (L2 → cosine on the generic LanceDB path, which was already cosine foragent_skilland on Milvus) is documented under Changed — it shifts ranking but requires no action, and the boilerplate## Upgradethe workflow appends carries the plain upgrade line.RC verification (against the package installed from PyPI)
Installed by exact pin (
everos==1.3.0rc1), both installation paths,everos init→ server → real interfaces:[milvus]extraeveralgo-boundary/httpxDetectionResult._fields('cells','tail')/add→/flush→ cascadeextracted, 0/0/0extracted, 0/0/0/get episode/searchdiscrimination (4 methods x 3 unrelated topics)All seven memory kinds produce rows on the published artifact:
episode6,atomic_fact80,foresight38,agent_case5,agent_skill5,user_profile1,knowledge_topic17.The single LanceDB miss is
keywordx one topic: that episode was extracted in Spanish from English input, so English BM25 terms cannot reach it. Known LLM instruction-following issue, unrelated to the index backend, absent from the Milvus run on the same input. Measured separately at 9/12 for that content and 0/12 for another — content-correlated, not random.