Neo4J Bolt Laravel Package

Product Decisions This Supports

• Graph-feature feasibility spikes. Answers the roadmap question "can our PHP/Laravel stack talk to Neo4j natively?" in days rather than weeks, before committing to a graph tier.
• Build vs buy for the graph data layer. This package is a raw Bolt wire-protocol driver, not an OGM/repository layer — evaluating it forces the team to price the "we still have to build models, migrations, transactions, and pooling ourselves" work that a managed client (e.g. laudis/neo4j-php-client) bundles.
• Roadmap sequencing. Informs whether Neo4j becomes a first-class dependency of the monolith, an isolated service behind an API, or stays a research-only tool.
• Use-case validation where graph traversal beats SQL joins: knowledge graphs, recommendations/social graphs, fraud-ring detection, data lineage/catalog, entitlement and permission graphs, supply-chain dependency mapping.
• Architecture lock-in review. Bolt is Neo4j's stable wire protocol, so a working PHP Bolt client proves PHP can be a first-class consumer of the graph tier rather than a second-class citizen behind a Python/Java sidecar.
• Not a signal for anything customer-facing: 1 star, 0 dependents, and no release since August 2020 means the market has effectively not voted. Absence of adoption is itself a decision input.

When to Consider This Package

Consider it when:

• You need a throwaway, in-process client for a time-boxed (1–2 week) feasibility spike, not a shipping dependency.
• You specifically need to speak Bolt directly — protocol conformance tests, proxy/instrumentation, load-generation tooling, or a custom client for an internal platform.
• Your Neo4j target is pinned to a 3.5/4.0-era server and won't move during the exercise.
• You intentionally want zero framework coupling — no service provider, no Eloquent-style models, no auto-discovery.
• You're prepared to vendor or fork it under MIT and name an owner for it on day one.

Look elsewhere when:

• The feature is production or customer-facing. Use laudis/neo4j-php-client (maintained, Bolt + HTTP transports, Symfony bundle and Laravel bindings), an official Neo4j driver in a sidecar service, or Neo4j's HTTP transactional endpoint for low-volume needs.
• You need Neo4j 5.x / Bolt 5, multi-database routing, bookmarks, managed (retryable) transactions, or modern auth schemes.
• You need connection pooling, backpressure/lazy result streaming, structured logging, or tracing.
• Your org requires documentation, security advisories, CVE response, or a support path — none of that exists here, and with zero dependents there is no ecosystem safety net absorbing breakage.
• Anyone on the team would be surprised to learn they own a protocol driver.

Gate to pass before adoption: a named maintainer, a written upgrade path onto a maintained client, and a scheduled review date. If any of the three is missing, this stays a spike.

How to Pitch It (Stakeholders)

Executives (one paragraph): This is a one-star, single-maintainer PHP driver for Neo4j's Bolt protocol whose last release was August 2020. It is a useful research instrument, not a platform dependency. The cost to try it is a few engineering days; the cost to own it is that we would effectively fork and maintain a graph protocol driver ourselves as Neo4j versions advance — a recurring tax with no competitive upside. Recommendation: fund a time-boxed spike (≤ 2 weeks, one engineer) to validate specific graph use cases, and keep it off the production roadmap. If the use case proves out, the follow-on investment is "adopt a maintained client or isolate Neo4j behind a service," not "maintain our own Bolt driver."

To engineering: MIT-licensed, small surface area, no framework coupling — genuinely useful for understanding the Bolt handshake and message flow, and as a reference implementation. But it predates Bolt 4.3/5.x, so expect to patch version negotiation, auth, TLS, and result streaming yourself, and to add the pooling and retry semantics it never had. Treat it as a starting point for a fork you own, never as a composer require in a product.

To finance/procurement: Zero licence cost. That is the least interesting number in the analysis — the total cost of ownership is the engineering time to keep a protocol driver current, and the risk sits entirely with us.

Framing the alternative: if the graph use case is real, the cheapest durable answer is usually a maintained PHP client or a thin graph service we call over HTTP. Both cost more to start and far less to keep — and neither makes our roadmap dependent on one repository with one star.