✦ This isn't a glossary anymore
A curated knowledge graph with editorial governance
A typical glossary gives you terms, definitions, and a search box. This goes further.
Terms are connected by reviewed, typed relationships, making it possible to follow how ideas relate across the entire corpus. Every connection is sourced, reviewed, and tracked, while the system continuously checks for missing links, dead references, and gaps in coverage.
The result is less a glossary and more a curated knowledge graph — designed not just to define concepts, but to show how they connect.
Honestly, we've stopped trying to pin down exactly what this is — and it's still being built. Depending on the day it reads like an Engineering Knowledge Operating System, a Curated Engineering Knowledge Graph, or an AI-Native Developer Knowledge Base, with a Blast Radius risk layer wiring in now. It's quietly becoming all of them — tap a ? to see what each means.
We're not quite sure what to call it — but it outgrew “glossary” a while ago.
The most ambitious framing: not a reference you read, but a substrate other tools run on. 819 reviewed, typed relationships bind the corpus into one coherent layer — alongside four-axis risk scores and editorial governance — that people browse and machines query. Engineering knowledge that's stored, connected, checked, and acted on all in the same place.
At its core it's a graph: 1,519 engineering concepts joined by reviewed, typed relationships. Every edge is sourced and tracked, every term scored on a four-axis risk model, and the whole corpus is continuously audited for missing links, dead references and gaps in coverage — closer to a maintained dataset than a wiki.
Built to be read by models as much as by people. The corpus is machine-readable end to end — typed edges, structured fields, an agent-sync layer, a deterministic assistant that answers from the graph instead of guessing, and 18 mediated narratives that walk through how distant concepts connect — so an LLM can traverse it the way a developer would.
The layer wiring in now. Every term already carries a four-axis risk score; Blast Radius teaches the graph to trace how that risk travels along its connections — turning isolated scores into a blast radius that shows how far a concept's risk actually reaches. Risk-graph readiness is at 40% and climbing as more connections are reviewed.
Glossary
A developer glossary, written to teach — built to be cited.
1.5k terms across 34 categories — definitions, examples, common mistakes and fixes. Authored under one editorial standard. Cross-linked, searchable, and openly crawled by AI agents.
Trending categories
AI citations · 7d
Popular tags
by term count
✦ Latest typed relationships
The graph is growing
Each edge is LLM-proposed, gate-filtered, and human-reviewed.
- URL & URLSearchParams API Often seen in AJAX Patterns for PHP Backends 1d
- URL & URLSearchParams API Often seen in Axios — HTTP Client for PHP APIs 1d
- Attack Surface Often seen in Penetration Testing 1d
- URL & URLSearchParams API Often seen in History API — SPA Routing with PHP Backend 1d
- URL & URLSearchParams API Often seen in Fetch API & HTTP Requests 1d
Garbage collection is the concrete piece of memory management that touches memory leaks directly: as a specialization of the heap-management side of the discipline, it's the runtime mechanism responsible for reclaiming unreachable objects. A leak is precisely what happens when that reclamation fails — most acutely in PHP when reference counting can't free circular references and the cycle collector is what stands between you and unbounded process growth. So the story from memory_management to memory_leak runs through GC because GC is the part of memory management whose success or failure determines whether allocations are ever returned.
Binary trees connect to pagination through database indexing, which is the structure that actually carries the benefit between them. The index is typically a B-tree variant of the binary tree concept, and its ordered, logarithmic-lookup structure is exactly what lets efficient pagination seek directly to a page's starting key (keyset pagination) instead of scanning and counting rows. Without the index sitting in the middle, the tree's ordering properties would never reach pagination at all — it's the index that translates tree structure into fast page boundaries.
Exactly-once delivery connects to domain events through the outbox pattern, which atomically writes those events to a database table in the same transaction as the business data that produced them. The pattern is what carries the events safely to the broker without loss or duplication, so the delivery guarantee is meaningfully about the domain events being relayed — remove the outbox and there's no mechanism tying the messaging guarantee to the architectural event records. That atomic write is the load-bearing piece: it's the bridge between 'something happened in the domain' and 'the broker delivers it exactly once'.
🤖 AI Guestbook educational data only
|
|
Last 30 days
Agents 351
Common Crawl 94
PetalBot 83
Google 76
SEMrush 27
Bing 25
ChatGPT 23
Brave Search 10
Ahrefs 8
Amazonbot 3
Applebot 2
Google 251
PetalBot 148
SEMrush 57
Bing 56
ChatGPT 42
Ahrefs 34
Brave Search 31
DuckDuckGo 4
Applebot 1
Amazonbot 1
ChatGPT 13k
Amazonbot 12.8k
Scrapy 11.7k
Perplexity 9.8k
Google 7.7k
Ahrefs 7.6k
SEMrush 6.1k
PetalBot 5.3k
Claude 3.4k
Unknown AI 3k
Bing 2.8k
Twitter/X 2k
Brave Search 1.9k
Meta AI 1.7k
Applebot 1.6k
Majestic 891
Sogou 439
Common Crawl 252
Qwen 190
You.com 99
ShapBot 36
DuckDuckGo 31
Yandex 15
NotebookLM 8
LinkedIn 2
Backlinks.com 1
Manus AI 1
Most referenced terms
APCu — In-Process User Cache 5
readonly Classes (PHP 8.2) 3
Maintainability Index 3
Psalm Type Annotations 3
Network Sockets 3
Abstract Classes 2
Reverse Proxy vs Load Balancer 2
allow_url_fopen / allow_url_include 2
Most active categories
How they use it
crawler 85.9k
crawler_json 5.8k
rag 1
pre-tracking
739
First ping
23 Mar 2026
Last 7 days
5.2k pings
Distinct agents
26
Unidentified
3k
2FA Bypass Techniques
PHP 5.0+
Common ways attackers circumvent two-factor authentication — SIM swapping, real-time phishing proxies, SS7 attacks, backup code theft, and session cookie hijacking after authentication.
d9
Security advanced
AI Context Management
PHP 8.0+
The practice of selecting, ordering, and trimming what goes into an LLM's context window to maximise relevance while staying under token limits.
d9
AI / ML intermediate
AI Model Quantization
Compressing neural network weights and activations to lower-precision formats (int8, int4, fp8) to shrink memory and accelerate inference.
d9
AI / ML advanced
AI Model Selection Criteria
The systematic factors engineers weigh when choosing an LLM for a task: capability, cost, latency, context window, modality, hosting, and licensing.
d9
AI / ML intermediate
AI Synthetic Data Generation
Using generative models to produce artificial training, testing, or augmentation data that mimics the statistical properties of real datasets without exposing originals.
d9
AI / ML intermediate
Apache Kafka Concepts
1
Kafka is a distributed log — producers append to immutable, ordered topics partitioned for parallelism, consumers read at their own pace and replay from any offset.
d9
Messaging intermediate
Apache Kafka Fundamentals
PHP 7.0+
A distributed event streaming platform that stores messages as an immutable ordered log partitioned across a cluster — optimised for high-throughput, durable, replayable event streams rather than traditional task queues.
d9
Messaging advanced
At-Least-Once Delivery
2
At-least-once delivery guarantees a message is delivered to at least one consumer, possibly multiple times — consumers must be idempotent to handle duplicates safely.
d9
Messaging intermediate
Authorisation
PHP 7.0+
4
The process of determining what an authenticated user is permitted to do — checking permissions, roles, or policies before allowing access to a resource or action.
d9
Security intermediate
BigInt — Arbitrary Precision Integers
ES2020
JavaScript's Number type cannot safely represent integers larger than 2⁵³−1 (Number.MAX_SAFE_INTEGER). BigInt is a separate numeric type that handles integers of arbitrary size, essential for working with 64-bit IDs from databases, cryptographic values, and precise financial calculations.
d9
JavaScript intermediate
Blameless Culture
An organisational norm where incidents and mistakes are treated as systemic learning opportunities rather than individual failures deserving punishment.
d9
General beginner
Blameless Post-Mortem
PHP 5.0+
3
A structured review of an incident focused on systemic causes and improvements, not individual blame — making it safe to surface failures.
d9
DevOps intermediate
Bloom Filter
PHP 7.0+ A probabilistic data structure that tests set membership in O(1) time and O(1) space, with a tunable false-positive rate and zero false negatives.
d9
Data Structures advanced
BM25 Ranking
Best Match 25 — the industry-standard relevance ranking algorithm used by Elasticsearch, Lucene, and SQLite FTS5, refining TF-IDF with better document length normalisation and a term frequency saturation parameter.
d9
Search intermediate
Boy Scout Rule
1
Always leave the codebase slightly cleaner than you found it — small, consistent improvements prevent entropy accumulation.
d9
Code Quality beginner
Brooks's Law
"Adding manpower to a late software project makes it later" — new developers require onboarding time and increase communication overhead before contributing productively.
d9
General intermediate
Bulkhead Pattern
Isolating system components into separate resource pools so a failure in one doesn't cascade and exhaust resources for others.
d9
Architecture advanced
Business Logic Abuse
Exploiting flaws in application workflows rather than technical vulnerabilities — bypassing payment steps, abusing discount codes, manipulating quantity fields, or racing concurrent requests.
d9
Security advanced
Business Logic Vulnerability
1
Flaws in application workflow allow attackers to abuse legitimate features in unintended ways.
CWE-840 OWASP A4:2021
d9
Security advanced
7.5
Cache Stampede / Thundering Herd
PHP 5.0+ When a cached item expires, multiple simultaneous requests all miss the cache and hit the database concurrently, overwhelming it.
d9
Performance advanced