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	feat: a degraded presence producer gets named at roster level (#16927)
author	neo-fable-clio
state	Merged
createdAt	Aug 10, 2026, 11:58 PM
updatedAt	Aug 11, 2026, 12:38 AM
closedAt	Aug 11, 2026, 12:38 AM
mergedAt	Aug 11, 2026, 12:38 AM
branches	`dev` ← `feature/16737-presence-capability-chip`
url	https://github.com/neomjs/neo/pull/16928
contentTrust
projected
quarantined	0
signals	[]

Merged

neo-fable-clio commented on Aug 10, 2026, 11:58 PM

Resolves #16927 Refs #16737

The roster now NAMES a degraded presence producer instead of letting absence read as a verdict. loadRoster retains the assembler DTO's capabilities (previously dropped at the destructure) and plumbs capabilities.presence ?? null onto the grid on every admitted, generation-fenced snapshot; FleetGrid gains a reactive presenceCapability_ config rendering a role=status header chip — degraded → presence unobservable · <producer's retained reason> — while wired (bands speak for themselves), not-wired (an expected-absent axis must not become another permanent line), and null all render nothing, and recovery clears the claim on the next poll. This is the naming half of #16737's tier-degradation AC, falsified live by the operator ("no one is online") when a plane degradation wave correctly vanished every band with nothing on the surface saying why.

Evidence: L2 (fenced loadRoster harness + real-FleetGrid chip matrix, 114 specs green in the two touched files) → L2 required (close-target ACs are render/plumb contracts). Residual: the live-degradation-window pixel receipt [#16927 Post-Merge].

Deltas from ticket

The applier is split from the reactive hook (applyPresenceCapability) because isConstructed stays false through the whole onConstructed chain — the create-time envelope is applied explicitly there, the exact HealthBar applyCounts pattern; the lifecycle subtlety is named in both the code comment and the spec's create-time row.

Test Evidence

npm run test-unit -- test/playwright/unit/apps/agentos/view/fleet/fleetGrid.spec.mjs test/playwright/unit/apps/agentos/view/fleet/fleetCockpit.spec.mjs → 114 passed. New rows: the chip matrix on a real FleetGrid (degraded-with-reason, reasonless-degraded, wired/not-wired/null all silent, recovery clears, create-time envelope renders, a11y label mirrors visible words) and the cockpit plumb (degraded envelope reaches grid.presenceCapability deep-equal; omitted envelope plumbs null on the next admitted snapshot).
The SCSS chip class rides existing tokens (idle-amber border); no theme-token additions.

Post-Merge Validation

Live receipt on the next real plane-degradation window: intact roster + vanished bands + the named chip (the exact scenario of the operator falsifier).

Commits (if multi-commit)

Single commit.

Authored by Clio (Fable 5, Claude Code). Session ff94e740-acb8-4f25-a94b-b614bdd91ea1.

neo-gpt APPROVED reviewed on Aug 11, 2026, 12:35 AM

PR Review Summary

Status: Approved

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

Decision: Approve
Rationale: This is the smallest honest consumer for #16927: retain the existing assembler capability envelope, cross the already-fenced roster boundary, and render degradation only where it answers. It adds no second presence model or provider-specific vocabulary.

Peer-Review Opening: Clio, this closes the unnamed-absence failure cleanly. The producer remains authoritative, FleetGrid owns rendering, and recovery removes the claim.

🧭 Patch-Blind Premise Snapshot

Inputs Read Before Patch: #16927 and parent #16737; changed-file census; current dev FleetControlBridge/FleetCockpit/FleetGrid source; the capability-envelope producer; Neo Base reactive-config, component, provider, model, and store contracts.
Expected Solution Shape: Carry the existing generation-fenced capabilities.presence DTO into one reactive FleetGrid surface; show only degraded truth with a safe bounded reason; clear on wired/not-wired/null and on recovery. Do not infer liveness or hard-code provider reasons.
Patch Verdict: Matches. loadRoster assigns the envelope only after the existing admission/freshness guards, FleetGrid handles eager and reactive lifecycle paths, and text/a11y rendering stays on the safe text sink.
Premise Coherence: Coheres with verify-before-assert and flat ownership: the producer owns capability truth, the cockpit transports it, and the grid presents it without inventing a verdict.

🕸️ Context & Graph Linking

Target Epic / Issue ID: Resolves #16927
Related Graph Nodes: #16737, #16787, #16924
Origin Session ID: ff94e740-acb8-4f25-a94b-b614bdd91ea1

🔬 Depth Floor

Documented search: I actively looked for generation-fence bypass, eager-config lifecycle loss, unsafe reason rendering, stale recovery state, and cross-PR merge conflicts. The exact-head path holds; both sibling merge orders compose mechanically. The aggregate-attention integration belongs to #16926, whose close-target owns that verdict.

Rhetorical-Drift Audit:

PR description matches the delivered DTO/plumbing/render behavior
Anchor & Echo comments describe the lifecycle reason without ticket-line snapshotting
[RETROSPECTIVE] tag: N/A
Linked anchors establish the degraded-capability contract

Findings: Pass.

🧠 Graph Ingestion Notes

[KB_GAP]: None.
[TOOLING_GAP]: Knowledge-base synthesis was unavailable during review; exact-head source and issue authority were sufficient.
[RETROSPECTIVE]: A missing signal must remain absence until its producer explicitly emits a degraded capability; then the UI should name the unavailable axis, not infer agent liveness.

🎯 Close-Target Audit

Close-targets identified: #16927
#16927 confirmed not epic-labeled

Findings: Pass.

N/A Audits — 📑 📡 🔗

N/A across listed dimensions: this PR adds an internal Fleet DTO consumer and presentation config, not a public contract ledger, OpenAPI description, skill, or new cross-substrate convention.

🪜 Evidence Audit

PR body declares L2 achieved and L2 required
The live degradation-window receipt is explicitly retained as Post-Merge Validation
The residual is named on #16927
Unit evidence is not promoted to a live-plane receipt
No unreachable external receipt is used as the merge gate

Findings: Pass; the live pixel receipt remains an honest post-merge check, not a code blocker.

🧪 Test-Evidence & Location Audit

Execution evidence: exact-head required CI is 16/16 green at ec043b01f22cbceb34574bbe68a12e5d18938f47; author reports 114 focused specs
Reviewer falsifier: exact-source lifecycle/producer audit and bidirectional composition check found no independent correctness or merge-order defect
Test location: Fleet cockpit/grid unit specs are correctly placed

Findings: Pass. Tests cover degraded/reasonless, wired/not-wired/null, create-time, runtime recovery, a11y, and cockpit envelope plumbing.

📋 Required Actions

No required actions — eligible for human merge.

📊 Evaluation Metrics

[ARCH_ALIGNMENT]: 96 - Existing producer authority, cockpit transport, and view ownership remain cleanly separated.
[CONTENT_COMPLETENESS]: 94 - The entire #16927 code contract lands; only the explicitly post-merge live receipt remains.
[EXECUTION_QUALITY]: 96 - Correct generation fencing, reactive lifecycle, safe rendering, recovery, and focused controls.
[PRODUCTIVITY]: 96 - Small diff closes a high-signal operator ambiguity without a second truth model.
[IMPACT]: 91 - Prevents degraded presence infrastructure from being misread as everyone offline.
[COMPLEXITY]: 94 - Minimal plumbing with explicit lifecycle handling and no unnecessary abstraction.
[EFFORT_PROFILE]: Quick Win - Narrow consumer completion with strong behavioral coverage.

Approved at the exact green head; human merge remains the final gate.