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	13538
title	Compress pull-request workflow payload
state	Closed
labels	documentationenhancementaiarchitecturemodel-experience
assignees	neo-gpt
createdAt	Jun 19, 2026, 11:27 AM
updatedAt	Jun 19, 2026, 1:57 PM
githubUrl	https://github.com/neomjs/neo/issues/13538
author	neo-gpt
commentsCount	1
parentIssue	10757
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[]
closedAt	Jun 19, 2026, 1:57 PM

Compress pull-request workflow payload

Closed v13.1.0/archive-v13-1-0-chunk-4 documentationenhancementaiarchitecturemodel-experience

neo-gpt commented on Jun 19, 2026, 11:27 AM

Context

This is the second focused reduction target after #13535 and #13537.

Current measured evidence:

node ./ai/scripts/lint/lint-skill-manifest.mjs --report-sizes --top 20 ranks .agents/skills/pull-request/references/pull-request-workflow.md as a top remaining hotspot: 34,096 bytes, 387 lines, 148 signals, disposition compress-to-trigger.
.agents/skills/skills.manifest.json still carries a temporary migration-period per-file budget override for pull-request (perFilePayloadBudget: 38000).
Discussion #11598 recorded pull-request-workflow.md at 36,041 bytes as part of the skill-payload accretion loop.
The pull-request workflow is hot because every authored PR must load it; bloat here directly taxes delivery and review routing.

Release classification: post-release Agent OS hardening; boardless.

The Problem

The pull-request workflow has accumulated branch discipline, commit hygiene, PR body, cross-family review, A2A routing, evidence declaration, authorship, and edge-case governance into one large map.

Those contracts matter, but their current shape is too expensive. The workflow also contains multiple edge-case blocks that should either be compact pointers to existing sibling payloads or compressed into smaller canonical rules. Continuing to add paragraphs to this file makes every PR handoff more expensive and increases the chance that smaller-context agents cannot complete a lifecycle without compaction.

The Architectural Reality

Primary target:

.agents/skills/pull-request/references/pull-request-workflow.md

Adjacent surfaces that must be checked but should not grow by default:

.agents/skills/pull-request/references/review-response-protocol.md
.agents/skills/pull-request/references/*
.agents/skills/skills.manifest.json

The pull-request skill is already a map with several sibling payloads. The corrective move is not to add a new book; it is to compress the hot workflow and use existing sibling payloads only when they reduce total invocation load.

The Fix

Perform a focused compression pass on pull-request-workflow.md.

Preferred direction:

Merge duplicated PR-open / handoff / A2A language.
Compress historical rationale into short source anchors.
Keep the required PR-body gates and cross-family review mandate grep-visible.
Avoid adding new sibling Markdown unless aggregate .agents/skills/pull-request/**/*.md bytes go down.
If the file drops enough, lower the temporary pull-request per-file manifest override in the same PR or name the follow-up target.

Contract Ledger Matrix

Target Surface	Source of Authority	Proposed Behavior	Fallback	Docs	Evidence
`.agents/skills/pull-request/references/pull-request-workflow.md`	`create-skill` recursive Map/Atlas guidance; #10757 cognitive-load audit	Reduce the hot PR-handoff payload while preserving required PR gates	Stop if compression weakens merge-safety or provenance contracts	Same workflow; PR body slot-rationale	before/after `wc -c`, `lint-skill-manifest --report-sizes --top 20`, grep for retained anchors
`.agents/skills/pull-request/references/**`	Existing sibling-payload model	Keep aggregate pull-request Markdown negative; no relocation-only split	Leave sibling payloads untouched if main-workflow compression is enough	Existing refs	aggregate `find .agents/skills/pull-request -name '*.md' -print0
`.agents/skills/skills.manifest.json`	Skill Manifest Lint	Lower temporary `pull-request` per-file override if the file safely fits a lower threshold	Document follow-up target if scope would be too broad	manifest	lint-skill-manifest

Decision Record impact

Aligned with ADR 0008 (SKILL.md anatomy / Progressive Disclosure) and the recursive workflow-map guidance in .agents/skills/create-skill/references/skill-authoring-guide.md.

Acceptance Criteria

Record before/after byte counts for .agents/skills/pull-request/references/pull-request-workflow.md.
Record before/after aggregate byte counts for .agents/skills/pull-request/**/*.md.
The PR has a net-negative .agents/skills/pull-request/**/*.md byte delta.
No new pull-request Markdown file is added unless aggregate pull-request Markdown remains negative and the PR explains why the split reduces invocation load.
Existing required PR contracts remain grep-verifiable: substrate-mutation pre-flight, ticket assignment gate, branch-before-code discipline, branch freshness, commit subject ticket ID, explicit --base dev, self-identification, Resolves-only PR body hygiene, Evidence declaration, cross-family mandate, A2A review routing, human-only merge, and authorship respect.
node ./ai/scripts/lint/lint-skill-manifest.mjs --base origin/dev passes.
If the resulting workflow fits below a lower per-file threshold, either lower the manifest override in the same PR or document the follow-up threshold target.

Out of Scope

Changing PR policy semantics.
Adding another pull-request workflow book.
Compressing pr-review or ideation-sandbox; those are separate hotspots.
Rewriting PR body lint or GitHub workflow APIs.

Avoided Traps

Relocation-only split: rejected because pull-request invocation would still pay the same aggregate cost.
Weakening merge gates: rejected; compression must preserve the safety contracts.
Adding more examples: rejected unless larger same-PR deletions make the aggregate negative.

Parent: #10757
Guard: #13533
First reduction target: #13535
Largest review target: #13537
Prior discussion anchor: #11598

Handoff Retrieval Hints

Search patterns:

pull-request-workflow skill bloat compression perFilePayloadBudget 38000
skill payloads accrete pull-request-workflow 36041
pull-request workflow net-negative map atlas