What is the Neural Link?

The Neural Link is a bi-directional bridge that connects AI agents directly to the Neo.mjs runtime. It lets agents inspect the Scene Graph, component state, event listeners, computed styles, and DOM rectangles, and mutate the running application in real time.

Why is Neo.mjs called an Application Engine instead of a framework?

Neo.mjs maintains persistent application objects in a worker-backed Scene Graph instead of compiling application state away into ephemeral DOM nodes. That architecture enables multi-window orchestration, runtime permutation, and deep AI introspection.

What is Context Engineering?

Context Engineering shapes the information and tool environment around AI agents. Neo.mjs implements it through Knowledge Base, Memory Core, GitHub Workflow, and Neural Link MCP servers for frontier harnesses, plus a File System MCP server for internal Neo.ai.Agent local loops.

What is the Neo.mjs Agent OS?

The Neo.mjs Agent OS is the repository Brain: source code and services for Memory Core, Knowledge Base, Active Hybrid GraphRAG, DreamService, Golden Path synthesis, A2A coordination, and Neural Link tooling.

Frontmatter

id	10304
title	Enrich manage_discussion_comment with Comment-ID hand-off protocol
state	Closed
labels	enhancementaiarchitecture
assignees	neo-gpt
createdAt	Apr 24, 2026, 7:37 PM
updatedAt	3:18 PM
githubUrl	https://github.com/neomjs/neo/issues/10304
author	neo-gemini-pro
commentsCount	0
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[]
blocking	[]
closedAt	Jun 5, 2026, 3:47 AM

Enrich manage_discussion_comment with Comment-ID hand-off protocol

Closed v13.0.0/archive-v13-0-0-chunk-6 enhancementaiarchitecture

neo-gemini-pro commented on Apr 24, 2026, 7:37 PM

Context

We implemented the A2A Comment-ID Hand-off Protocol for issues and PRs in #10272 / PR #10287. Ideation Sandbox Discussions can also grow long over multiple review cycles, so Discussion comments need the same linear-cost handoff shape.

Current State (2026-06-03 V-B-A)

manage_discussion_comment already returns commentId, url, and createdAt on create, plus commentId, url, and updatedAt on update. That slice was delivered by #10841 / PR #10891.
get_conversation supports comment_id, since_comment_id, and last_n selectors for Issues and Pull Requests only.
No get_discussion_conversation or equivalent Discussion selective-fetch read tool exists in the GitHub Workflow MCP surface.
ideation-sandbox discussion cycles currently still rely on full Discussion context or raw GraphQL/local artifacts, so the N^2 fetch problem remains for long design threads.

Problem

The remaining gap is Discussion-domain selective fetch. A2A messages can carry a Discussion commentId, but there is no agent-callable read tool that fetches only that comment or the comments after it.

Contract Ledger Matrix

Target Surface	Source of Authority	Proposed Behavior	Fallback / Edge Case	Docs	Evidence
`manage_discussion_comment` return payload	#10841 / PR #10891; current `DiscussionService.manageDiscussionComment()`	Already returns `commentId` + URL + timestamp. No new implementation required for this row.	Update path returns `updatedAt`, not `createdAt`; callers should treat timestamp field according to action.	`GitHubWorkflow.md`, OpenAPI if wording is stale	Source inspection + focused existing tests if touched
New Discussion selective-fetch tool	#10304; issue/PR `get_conversation` selector contract	Add `get_discussion_conversation({discussion_number, comment_id?, since_comment_id?, last_n?})` or an equivalent MCP read surface. Selector precedence mirrors issues/PRs: `comment_id > since_comment_id > last_n > full`.	Unknown `comment_id` returns empty comments; unknown `since_comment_id` returns empty comments; invalid or missing `discussion_number` returns structured error; full fetch remains the cold-cache fallback.	OpenAPI + `GitHubWorkflow.md`; `ideation-sandbox` handoff guidance if changed	Unit coverage for selector precedence and structured errors
Discussion comment shape	GitHub Discussion GraphQL schema + existing sync query conventions	Returned discussion includes title/body/author/category metadata plus filtered `comments.nodes`; comment nodes include `id`, `author.login`, `body`, `createdAt`, `isAnswer`, and replies if included by the existing query shape.	Replies remain nested under the returned parent comment; no separate reply selector in this ticket.	OpenAPI schema / docs	Service test with representative GraphQL fixture
Ideation Sandbox handoff usage	`ideation-sandbox` workflow + A2A commentId handoff precedent from `pr-review`	Long Discussion cycles can pass `commentId` via A2A so peers fetch only the new Discussion comment when warm-cache context exists.	Fresh/cold-cache reviewers still fetch broader context before evaluating a single isolated comment.	`ideation-sandbox` docs; optional cross-reference to `pr-review` warm-vs-cold discipline	Docs grep + review evidence

Acceptance Criteria

manage_discussion_comment returns commentId and timestamp metadata. Delivered by #10841 / PR #10891; no new code required.
A Discussion selective-fetch read tool exists with comment_id, since_comment_id, and last_n selectors.
Selector precedence and fallback semantics match issue/PR get_conversation behavior.
OpenAPI/tool documentation names the Discussion selector contract.
ideation-sandbox guidance documents commentId handoff for warm-cache Discussion cycles, with cold-cache fallback.
Focused tests cover the service/router selector behavior.

#10272 / PR #10287 — issue/PR comment-ID return + selective fetch.
#10841 / PR #10891 — Discussion commentId return payload.
#10376 / PR #10377 — A2A commentId handoff discipline; adjacent but does not close this Discussion-domain read-tool gap.

Origin Session ID: 0b29a8fa-c6b0-42e2-ab3b-8015a99db2d8

tobiu closed this issue on Jun 5, 2026, 3:47 AM

tobiu referenced in commit 2938379 - "feat(github-workflow): add discussion selective fetch (#10304) (#12552) on Jun 5, 2026, 3:47 AM