docs: address review feedback on SDK v5/v6 migration guides

[hassiebp]

Applies review feedback from #4008 to the Python v4 → v5 and JS/TS v5 → v6 migration guides and to the changelog post. All anchors stay the same. Each edit replaces or deletes text inside an existing section.

Both guides

• Intro: two hedged paragraphs become one precise paragraph that says what the major removes, what it needs, and where to check whether you're affected.
• "Are you affected?": unsupported self-hosted versions get their own bullet that says so directly.
• Minimum versions: the "stay on SDK v4/v5" sentence becomes a callout linking the Langfuse v3 → v4 upgrade guide.
• #scores-api heading: now "api.scores.create takes a list/array of scores".
• Internal type names removed from the prose: CreateScoreBatchResponse/CreateScoreBatchResults, GetScoresV3Request/GetScoresV3Response, and ObservationV2 (the Python batched-evaluation code comment keeps it). CreateScoreRequest stays because users import it.
• Score reads are described by method name (scores_v3.get_many_v3 → scores.get_many, score_ids → id) instead of "Scores API v2/v3".
• The paragraph listing removed request/response types is deleted.

Python guide: factual fix. v4 API deprecations use typing_extensions.deprecated(..., category=None), which emits no runtime warning, and the trace IO setters are deprecated only in their docstrings. The guide no longer claims these raised DeprecationWarning and points to type checkers instead. blocked_instrumentation_scopes, which does warn at runtime, is unchanged.

Changelog

• The intro now says the majors mostly remove already-deprecated APIs, so coding agents default to the current APIs.
• The self-host sentence links the Langfuse v4 upgrade guide, and the duplicated deprecation sentence is removed.

Relation to #4042: this branch merges in #4042 (checklists as plain bullets), with this PR's wording on the api.scores.create checklist item. It merges cleanly whether or not #4042 merges first.

Verification

• pnpm run format:check: "All matched files use Prettier code style!"
• node scripts/check-h1-headings.js: success.
• Dev server: all three pages return 200 with the new headings and callouts rendered and no MDX errors.

  

Confidence Score: 4/5

The PR appears safe to merge, with a non-blocking correction to the Python intro’s warning advice.

Summary

This PR makes the SDK migration guides clearer and updates the changelog.

• The guides show which SDK upgrades fit each server version and where to start.
• The guides explain score creation and reads through the public methods.
• The Python guide points readers to type checkers for deprecated calls.

hassiebp explicitly describes the removal of type-name documentation as intentional in the PR description.

_{Reviews (1) · Last reviewed commit: "docs: address review feedback on SDK v5/..." · Reviewed by Greptile}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
langfuse-docs	Ready	Preview	Oct 14, 2026 11:21am UTC

[github-actions]

@claude review

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-14T11:19:53.431055Z	d69ac17	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[claude]

Claude Code Review

No review was started: this request came from a bot account. Manual reviews can only be requested by someone with write access to this repository. Ask a maintainer to comment @claude review, or have your automation post the comment from a user account with write access.

_{Tip: disable this comment in your organization's Code Review settings.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d69ac17316

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[claude]

Nothing blocking. The comments below are optional suggestions. There is no need to push a fix for them before merging.

[claude]

🟡 nit (optional): "current, fastest APIs" in the new changelog intro is a vague, unsupported superlative, which the changelog guidelines in CLAUDE.md ask to avoid. Fix: drop "fastest" or name the concrete benefit (for example, the v2 observations and v3 scores endpoints).

Why this was flagged

The diff adds "so the SDKs expose only the current, fastest APIs" at 2026-10-14-python-sdk-v5-js-sdk-v6.mdx:13. CLAUDE.md's changelog rules say to be specific, avoid vague adjectives, and keep a non-marketing tone. Nothing in the entry or the linked guides backs up "fastest" with a measurement or a named endpoint. The base intro had no such claim.

Verification: nit. The diff adds a sentence at content/changelog/2026-10-14-python-sdk-v5-js-sdk-v6.mdx:13 saying the SDKs expose "only the current, fastest APIs". The base intro had no such claim. .agents/AGENTS.md line 129 says "Be specific — name actual filters, metrics, commands. Avoid vague adjectives." Nothing in the entry backs up "fastest". The impact is limited to copy quality.