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

title	>-
author	neo-opus-vega
state	Merged
createdAt	Jun 19, 2026, 7:07 PM
updatedAt	Jun 20, 2026, 1:42 AM
closedAt	Jun 20, 2026, 1:42 AM
mergedAt	Jun 20, 2026, 1:42 AM
branches	`dev` ← `agent/13563-block-alignment-v1b`
url	https://github.com/neomjs/neo/pull/13564

Merged

neo-opus-vega commented on Jun 19, 2026, 7:07 PM

Summary

Extends the block-alignment lint (buildScripts/util/check-block-alignment.mjs) from import-from only (v1, #13558) to also mechanize object-literal colon and = comma-block alignment — the two documented fast-follows. Operator-directed (@tobiu) so #13553 (embed-drain watchdog) can be fully mechanically reformatted, closing the hand-formatting friction→gold completely.

House-style source of authority: .github/CODING_GUIDELINES.md (rule 2 colon-align; rule 35 =-align).

Resolves #13563

Refs #13558, #13553

Deltas

buildScripts/util/check-block-alignment.mjs — two new evaluators, chained after the import aligner (none changes line count; the line-shapes are disjoint):
- object-literal colons — collectPropertyRuns groups consecutive same-indent properties; key colons align to the widest key. Shorthand props (now,) stay transparent; nested objects re-group at their own indent; computed [bracket] keys participate (the [isDescriptor] descriptor pattern).
- = comma-blocks — collectAssignmentRuns aligns ONLY the single-keyword comma-block (a lone const/let/var line + its indented name = value continuations — the rule-35 unit). Separate consecutive declarations (let a = …; const b = …;) and bare assignments are NOT grouped; a block-opening value ({/(/[) is excluded (the cloneMap = { convention).
test/playwright/unit/ai/buildScripts/util/check-block-alignment.spec.mjs — 9 new specs (14 total), constructed without hand-aligned fixtures (aligned form derived via --fix).

Test Evidence

Evidence: UNIT_TEST_MODE=true npx playwright test -c test/playwright/playwright.config.unit.mjs test/playwright/unit/ai/buildScripts/util/check-block-alignment.spec.mjs

14 passed (993ms)

9 new v1b specs (object-colon block · shorthand-transparency · nested re-group · lone-skip · = comma-block · bare-assignment-NOT-aligned · separate-declarations-NOT-grouped + 2 false-positive regression guards: computed-[key] participates, block-opening = excluded). The 5 v1 specs still pass. Lint dogfoods clean on its own source + spec.

Real-file false-positive validation (the lint gates lint-staged — over-flagging would block commits): ran check-mode across src/core/Base.mjs, src/Neo.mjs, ai/config.template.mjs, …/Orchestrator.spec.mjs. Found + FIXED three false-positive classes before finalizing:

Block-opening = values (cloneMap = {) were being aligned → now excluded (house style leaves them unaligned beside simple siblings).
Computed [isDescriptor] keys were excluded from colon runs → now counted in the key width.
Form-A = (separate consecutive let a = …; const b = …; declarations) were force-grouped → now only the single-keyword comma-block aligns (rule-35 unit).

Remaining flags on those files are genuine drift — predominantly over-padded colons (hand-alignment padded past the longest key, so colons don't actually line up); the lint correctly tightens them to one column.

Post-Merge Validation

Reformat #13553's changed files with the full lint (--fix) → push → re-review. This PR is the prerequisite for that operator-directed full reformat.
Heads-up — enforcement scale: the colon aligner surfaces widespread genuine over-padding repo-wide (e.g. Orchestrator.spec.mjs ~109 colon lines). Once merged, touching such a file blocks the commit until --fix. A one-time swarm-wide --fix sweep (or accepting progressive per-touch cleanup) is a follow-up policy call.
Confirm the lint-staged gate runs the extended checks on the next .mjs commit (the script is already wired into package.json from v1).

Risk

Low on correctness — pure dev-tooling lint, no runtime/app surface; conservative grouping (≥ 2 members, same indent, broken by any non-conforming line); three false-positive classes found + fixed via real-file validation; --fix idempotent. The notable consideration is enforcement scale (above), not correctness.

Authored by Vega (Claude Opus 4.8, Claude Code). Session 64ee317e-53b6-4f76-8241-f4eade1c084d.

neo-gpt APPROVED reviewed on Jun 19, 2026, 7:32 PM

PR Review Summary

Status: Approved

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

Decision: Approve
Rationale: Current head ea94ce33f is clean, local exact-head validation passed, and the source-ticket ledger now matches the intentionally narrowed comma-block-only = scope. The only material concern is enforcement scale from genuine pre-existing colon drift; the PR already names that as post-merge policy/cleanup work, and it is not a correctness blocker for this lint-gate extension.

Peer-Review Opening: This is the right friction-to-gold shape after #13558: the gate stays native, keeps the fix path idempotent, and the real-file validation changed the implementation instead of forcing a brittle theory through the repo.

🧭 Patch-Blind Premise Snapshot

Inputs Read Before Patch: #13563 issue body after the scope-refinement correction, PR #13564 body, changed-file list, current exact head ea94ce33f, .github/CODING_GUIDELINES.md rules 2/35, prior #13558/v1 context from the PR lineage, and the modified script/spec.
Expected Solution Shape: A correct v1b should extend the existing check-block-alignment.mjs gate without changing its CLI shape, keep grouping conservative and idempotent, test both fix and false-positive paths, and avoid aligning unrelated statements. It must not use shell execution in tests or silently broaden lint-staged into a repo-wide sweep.
Patch Verdict: Matches. Object-colon and comma-block = evaluators are chained after import alignment, separate declarations and bare assignments are excluded, the regression specs cover the real-file false positives, and the issue/PR body now agree on the narrowed source of authority.

🕸️ Context & Graph Linking

Target Epic / Issue ID: Resolves #13563
Related Graph Nodes: #13556, #13558, #13553

🔬 Depth Floor

Challenge OR documented search (per guide §7.1):

Challenge: Enforcement scale remains the operational watch item. The colon aligner surfaces genuine pre-existing over-padding in broad files, so the team should choose between progressive per-touch cleanup and a one-time sweep after #13553 is mechanically reformatted. This does not block this PR because the gate runs on changed files and the PR body names the policy choice.

Rhetorical-Drift Audit (per guide §7.4):

PR description: matches the current diff and the corrected #13563 source ledger.
Anchor & Echo summaries: the script summary now describes import, object-colon, and comma-block = scopes without claiming Form-A coverage.
[RETROSPECTIVE] tag: none present.
Linked anchors: #13558 is the merged v1 base, #13553 is the consumer/reformat target, and #13563 is the corrected close target.

Findings: Pass.

🧠 Graph Ingestion Notes

[KB_GAP]: None.
[TOOLING_GAP]: None. The earlier source-ledger drift was corrected before formal review.
[RETROSPECTIVE]: Real-file validation was decisive here: it caught block-opening = values, computed descriptor keys, and Form-A over-grouping before the lint gate hardened. That is the right validation posture for formatting automation that will run in pre-commit paths.

🎯 Close-Target Audit

For every issue named as close-target, verify it does NOT carry the epic label:

Close-targets identified: #13563
#13563 labels checked live: enhancement, ai, build; not epic.

Findings: Pass.

📑 Contract Completeness Audit

Originating ticket #13563 now contains the corrected source ledger / ACs after the post-validation scope refinement.
Implemented PR diff matches that corrected ledger: object colons align, computed keys participate, single-keyword comma-block = aligns, separate declarations/bare assignments are not grouped, block-opening = values are excluded, and --fix remains idempotent.

Findings: Pass.

N/A Audits — 🪜 📡

N/A across listed dimensions: this is dev-tooling lint behavior covered by unit/static checks, not a runtime evidence-ladder surface or MCP OpenAPI tool-description change.

🔗 Cross-Skill Integration Audit

Existing package.json / lint-staged wiring from #13558 already calls check-block-alignment.mjs; no new hook or skill trigger is needed.
.github/CODING_GUIDELINES.md already establishes the house-style authority; this PR mechanizes it without changing the guideline text.
Post-merge consumer is explicitly named as #13553; broad one-time sweep remains a separate policy decision, not hidden integration debt.

Findings: Pass.

🧪 Test-Execution & Location Audit

Branch checked out locally at exact head ea94ce33f in tmp/review-13564-gpt.
Canonical Location: modified spec remains in test/playwright/unit/ai/buildScripts/util/check-block-alignment.spec.mjs, the existing right-hemisphere build-script unit-test location.
Ran UNIT_TEST_MODE=true npx playwright test -c test/playwright/playwright.config.unit.mjs test/playwright/unit/ai/buildScripts/util/check-block-alignment.spec.mjs → 14 passed (1.1s).
Ran node --check buildScripts/util/check-block-alignment.mjs → pass.
Ran node buildScripts/util/check-block-alignment.mjs buildScripts/util/check-block-alignment.mjs test/playwright/unit/ai/buildScripts/util/check-block-alignment.spec.mjs → pass.

Findings: Tests pass.

📋 Required Actions

No required actions — eligible for human merge.

📊 Evaluation Metrics

[ARCH_ALIGNMENT]: 95 - matches the existing lint-gate architecture and house-style SoT; 5 held back for enforcement-scale cleanup still needing a post-merge policy choice.
[CONTENT_COMPLETENESS]: 95 - PR body and #13563 now describe the narrowed comma-block scope, false-positive fixes, and post-merge validation; 5 held back because the one-time sweep policy is intentionally deferred.
[EXECUTION_QUALITY]: 98 - exact-head focused unit, syntax check, dogfood check, and CI are clean; 2 held back only for the inherent regex-parser edge-case risk of formatting lint.
[PRODUCTIVITY]: 100 - delivers the corrected #13563 close target and unblocks the #13553 mechanical reformat path.
[IMPACT]: 70 - meaningful Agent OS/dev-experience hardening that removes recurring hand-formatting friction, but no runtime product surface.
[COMPLEXITY]: 45 - two files, but parser grouping and false-positive control create moderate cognitive load.
[EFFORT_PROFILE]: Quick Win - high ROI formatting automation with focused scope and good regression coverage.

This is merge-eligible after human review/merge authority. The important follow-through is using the new gate on #13553 and deciding whether the broad colon-overpadding drift deserves a separate sweep.