[AIT-206] Add message per token guide for LangGraph

[matt423]

Description

Follows the existing structure of the message-per-token guides for LangGraph in JS.

Review App

Checklist

• Commits have been rebased.
• Linting has been run against the changed file(s).
• The PR adheres to the writing style guide and contribution guide.

Summary by CodeRabbit

• Documentation
  • Added new guides for token streaming implementations with Vercel AI SDK (message-per-response) and LangGraph (message-per-token) approaches
  • Included comprehensive LangGraph tutorial with prerequisites, setup instructions, code examples, event sequencing patterns, and best practices for integration

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Walkthrough

This PR adds documentation and navigation for a new guide on token-by-token streaming with LangGraph. It includes navigation entries, an index tile linking to the guide, and a comprehensive tutorial document with publisher and subscriber code examples using Ably.

Changes

Cohort / File(s)	Summary
Navigation and Index Updates src/data/nav/aitransport.ts, src/pages/docs/ai-transport/index.mdx	Added two new navigation entries for AI transport guides: Vercel AI SDK message-per-response and LangGraph message-per-token. Updated the AI Transport index to include a new LangGraph section with a tile linking to the message-per-token guide.
Documentation src/pages/docs/guides/ai-transport/lang-graph-message-per-token.mdx	New comprehensive guide documenting token-by-token streaming with LangGraph over Ably. Includes prerequisites, setup instructions, code examples for publisher and subscriber implementations, event sequencing details, and notes on throughput and echo prevention.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~22 minutes

Poem

🐰 Through LangGraph's token streams we hop,
Each message dances, won't stop!
With Ably's flow, granular and bright,
A guide to make streaming right. 📚✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The PR title accurately describes the main change: adding a message-per-token guide for LangGraph, which is clearly reflected in the code changes.
Linked Issues check	✅ Passed	The PR implementation fulfills the AIT-206 objective by creating a comprehensive guide article demonstrating token streaming with granular history using LangGraph in JavaScript.
Out of Scope Changes check	✅ Passed	All changes are directly related to AIT-206 objectives: new navigation entries, updated index documentation, and the new LangGraph guide file are all within scope.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[matt423]

@coderabbitai review

[coderabbitai]

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

Actionable comments posted: 3

🤖 Fix all issues with AI agents

In `@src/data/nav/aitransport.ts`:
- Around line 107-114: The nav entry named 'Vercel AI SDK token streaming -
message per response' points to a non-existent page at
'/docs/guides/ai-transport/vercel-message-per-response'; either create the
missing guide file or update the nav link to the existing Vercel guide path
'/docs/guides/ai-transport/vercel-message-per-token' (or rename the entry to
match the existing file) by editing the object in src/data/nav/aitransport.ts so
the name and link correspond to the actual MDX file.

In `@src/pages/docs/guides/ai-transport/lang-graph-message-per-token.mdx`:
- Around line 268-296: The responses Map is never cleaned up, causing unbounded
growth; in the stop handler (the channel.subscribe('stop', ...) callback where
responseId and finalText are obtained) delete the entry from responses after you
finish using finalText (e.g., call responses.delete(responseId)) so per-response
state is released when the response completes.
- Around line 180-226: The module-scoped responseId causes cross-request reuse;
make responseId a local variable inside streamLangGraphResponse (declare let
responseId = null at the top of that function) so each call gets its own ID,
update where you check and set it (inside the for-await loop) and include it in
the start and token publishes as before, and guard the final stop publish so you
only call channel.publish({ name: 'stop', ... }) when responseId was captured
(i.e., if (responseId) publish stop) to avoid emitting stop for unrelated
streams.