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.

Neo.mjs - Self-Evolving Software Organism

Frontmatter

title	fix(ai): bound LMS REM graph output (#13981)
author	neo-gpt
state	Closed
createdAt	Jun 25, 2026, 3:08 AM
updatedAt	Jul 27, 2026, 12:07 AM
closedAt	Jun 25, 2026, 3:30 AM
mergedAt
branches	`dev` ← `codex/13981-lms-rem-output-cap`
url	https://github.com/neomjs/neo/pull/13982
contentTrust
projected
quarantined	0
signals	[]

Closed

neo-gpt commented on Jun 25, 2026, 3:08 AM

Resolves #13981

Bounds REM graph structured-output generation for LMS/OpenAI-compatible chat calls so turn-boundary input chunking is not undermined by unbounded JSON output. The patch adds an ADR-19 AiConfig.localModels.chat.graphOutputLimitTokens leaf, applies it to Tri-Vector and topology graph extraction calls via max_tokens, preserves LMS streaming finish_reason metadata when present, and logs chunk-start details before each Tri-Vector provider call.

Evidence: L2 unit/static validation covers #13981 AC1-AC5. Residual: post-merge live REM backlog drain should confirm LMS no longer grows past the bounded chunk/output contract on the active backlog.

Deltas from ticket

Used max_tokens only for LMS/OpenAI-compatible payload compatibility; no wall-clock timeout was added.
Kept provider envelope compatibility: finish_reason fields are only added when LMS sends finish metadata.
Applied the same graph output cap to topology inference because it is the second REM graph-provider call path.

Test Evidence

node --check ai/config.template.mjs
node --check ai/provider/OpenAiCompatible.mjs
node --check ai/services/graph/TopologyInferenceEngine.mjs
git diff --check
npm run test-unit -- test/playwright/unit/ai/provider/KeepAlive.spec.mjs — 22 passed, 1 skipped
npm run test-unit -- test/playwright/unit/ai/services/graph/SemanticGraphExtractor.spec.mjs — 12 passed
npm run test-unit -- test/playwright/unit/ai/config.template.spec.mjs — 8 passed
npm run agent-preflight -- ai/config.template.mjs ai/provider/OpenAiCompatible.mjs ai/services/graph/SemanticGraphExtractor.mjs ai/services/graph/TopologyInferenceEngine.mjs test/playwright/unit/ai/config.template.spec.mjs test/playwright/unit/ai/provider/KeepAlive.spec.mjs test/playwright/unit/ai/services/graph/SemanticGraphExtractor.spec.mjs

Post-Merge Validation

Pull dev, run the config overlay update script, restart the orchestrator, and confirm the pending REM backlog resumes with LMS chunk logs and without over-cap growth.

Commit

079858ee28 — fix(ai): bound LMS REM graph output (#13981)

Authored by Euclid (GPT-5, Codex Desktop). Session 019ef378-527d-7393-bc74-ec3a1d3f2ddf.

tobiu commented on Jun 25, 2026, 3:30 AM

REJECTED. this does not help. and 8k tokens is WAY too low.