Permanent Notes
Last updated: 2026-06-27 (operation: setup)
| Note | Domain | Summary |
|---|---|---|
| Concept Name | # permanent/ — Atomic Concept Notes | |
| Action Environment Gap | The Action Environment Gap is the difference betwe | |
| Agent State Checkpointing | Agent State Checkpointing is a technique used to e | |
| Agent Substitutability | Agent Substitutability is a design principle in mu | |
| Agent Tool Scoping | # Agent Tool Scoping | |
| Agentic Loop | The agentic loop is the core control-flow pattern | |
| Agentic Systems | Agentic systems are systems that can autonomously | |
| Agile Cadence | Cadence trong Agile là nhịp điệu cố định của các s | |
| Agile Manifesto | Agile Manifesto (2001) là tuyên ngôn của 17 practi | |
| AI-Assisted Testing | AI-assisted testing là cách tiếp cận kết hợp AI co | |
| Ambiguous Goal | An ambiguous goal is a high-level objective that l | |
| Assume and Process | Assume and Process is a strategy for handling ambi | |
| Audit Logging for Agent Handoffs | Comprehensive audit logging is essential for track | |
| Authority Boundaries | Authority Boundaries define the scope of a sub-age | |
| AWS Lambda Cold Start | Every Lambda invocation goes through two phases. T | |
| CALMR | CALMR là mô hình 5 trụ cột của [[permanent/devops| | | |
| | Cassandra High Availability on Kubernetes with Spot Instances | Cassandra’s own HA model — no single point of fail | |
| CCAF Framework | The CCAF (Claude Certified Architect Framework) is | |
| Chatbots | Chatbots are systems that can generate human-like | |
| China Data Regulation | Trung Quốc xây dựng chế độ quản lý dữ liệu trên ** | |
| Continuous Delivery | CI/CD là cặp khái niệm thường đi cùng nhau, tạo th | |
| CIP và Counterpart Fund — Lương QLVNCH Ẩn Trong Viện Trợ Kinh Tế | - Commercial Import Program | |
| Clarify First Strategy | The Clarify First strategy is a cautious approach | |
| Container Runtime Security with Falco | [“Falco”, “runtime security”, “container runtime m | |
| Context Capacity | Context Capacity refers to the amount of informati | |
| Context Isolation | Context Isolation is a design principle where sub- | |
| Continuous Delivery (CD) | Continuous Delivery là thực hành đảm bảo rằng code | |
| Continuous Deployment | Continuous Deployment là bước tiến xa nhất trong c | |
| Continuous Integration (CI) | Continuous Integration là thực hành tự động hóa tấ | |
| Continuous Learning Culture (CLC) | Continuous Learning Culture (CLC) là một trong 7 n | |
| Cost of Delay | Cost of Delay (CoD) là giá trị kinh tế bị mất đi d | |
| Danh xưng làng Bắc Bộ qua các thời kỳ | Tên làng Việt Bắc Bộ là một lớp trầm tích lịch sử | |
| Decomposed Task | A decomposed task is a sub-task that has been brok | |
| Dependency Graph (Agentic) | In agentic systems, a Dependency Graph is a direct | |
| Deterministic Workflows | Deterministic workflows are fixed sequences of ste | |
| DevOps Pipeline Stages | Một DevOps pipeline đầy đủ gồm 9 stage liên tiếp, | |
| DevOps Topology | DevOps Topology là framework phân loại các cách tổ | |
| DevOps Transformation Canvas | DevOps Transformation Canvas là công cụ workshop t | |
| DevOps | DevOps là tập hợp các thực hành kết hợp phát triển | |
| DORA Metrics | [ | |
| Dynamic Planning | Dynamic planning is the ability of an agent to ada | |
| Dynamic Replanning | Dynamic replanning is the process of revising a pl | |
| Environment Errors | Environment Errors manifest as failures in the und | |
| Error Cascade | An Error Cascade is a critical failure mode in mul | |
| Error Classification Framework | An Error Classification Framework is essential for | |
| Error Propagation in Multi-Agent Systems | In multi-agent systems, it is vital for any failur | |
| Error Threshold | An Error Threshold is a form of loop control where | |
| Escalation Trigger | Escalation Triggers are explicit programmatic cond | |
| Escalation (Agentic) | Escalation is a loop control mechanism where an ag | |
| Evaluator Optimizer | The Evaluator Optimizer is a workflow pattern that | |
| Fallback Chains | A Fallback Chain is a systematic strategy for mana | |
| False Parallelism | False parallelism is an anti-pattern in agentic ar | |
| Fan-in (Agentic) | In agentic systems, Fan-in is the crucial sync | |
| Fan-out (Agentic) | In agentic systems, Fan-out is the process wit | |
| Feature Flag | Feature flag là kỹ thuật cho phép bật/tắt một tính | |
| Flow Framework | Flow Framework là framework của Mik Kersten (tác g | |
| Four-Phase Lifecycle of a Tool Call | The four-phase lifecycle of a tool call describes | |
| Git Object Model | Git stores everything as objects in `.git/obje | |
| Goal Drift | Goal Drift is a significant risk in adaptive plann | |
| Graceful Degradation | Graceful Degradation is the strategy of returning | |
| Gươm, Kiếm, Đao — Vietnamese Sword Taxonomy | # Gươm, Kiếm, Đao — Vietnamese Sword Taxonomy | |
| Handoff Messages | Handoff Messages are structured payloads passed be | |
| Handoff Point | A Handoff Point is the critical boundary in a mult | |
| Handoff Protocol | A handoff protocol is a formal contract governing | |
| Handoff Verification | Handoff Verification is a process that ensures a s | |
| Hard Block vs Soft Warning | Enforcement patterns for validation gates depend o | |
| Hedged Requests | Hedged Requests là kỹ thuật client-side giảm thiểu | |
| Hierarchical Decomposition | Hierarchical Decomposition is a task decomposition | |
| Hub-and-Spoke Topology | The Hub-and-Spoke topology is a multi-agent orches | |
| Human-in-the-Loop (HITL) | Human-in-the-Loop (HITL) is a deliberate architect | |
| Hybrid Topologies | Hybrid topologies in multi-agent orchestration com | |
| Incremental Complexity | Incremental Complexity is a design principle that | |
| Independent Testability | Independent testability is a characteristic of a w | |
| Infinite Loops | Infinite loops are a failure mode in agentic syste | |
| Instruction Design Principles | Instruction Design Principles for sub-agents state | |
| Interruption Point | An Interruption Point is a defined location in an | |
| Iterative Refinement Loops | Iterative Refinement Loops, also known as generate | |
| K-Means Clustering | K-Means Clustering là thuật toán phân cụm cơ bản n | |
| Kingman’s Formula | Kingman’s Formula (còn gọi là VUT equation) là côn | |
| Kolla-Ansible Deployment Patterns | Kolla-Ansible đơn giản hóa việc deploy OpenStack b | |
| Kotter’s 8-Step Change Model | John Kotter đề xuất mô hình 8 bước để dẫn dắt tran | |
| Kubernetes Node Consolidation | Node consolidation là chiến lược phân phối workloa | |
| Kubernetes Node Pool Design | Node pool design là việc tổ chức các node trong Ku | |
| Lầu Năm Góc và Khoảng Trống Giám Sát Của Quốc Hội Mỹ Trong Chiến Tranh Vietnam | [ | |
| Layered Enforcement Architecture | Layered Enforcement Architecture is a defense-in-d | |
| Little’s Law | Little’s Law là định lý của John Little, mô tả mối | |
| LLM Wiki Pattern | LLM Wiki là một pattern để xây dựng personal knowl | |
| Lossless Semantic Tree | Lossless Semantic Tree (LST) là một phương pháp ph | |
| Machine Learning | Machine Learning là tập con của AI, bao gồm các th | |
| MAP và EDA — Định Giá Thấp Vũ Khí Viện Trợ cho QLVNCH | [ | |
| MCP Architecture | # MCP Architecture | |
| MCP Primitives | The Model Context Protocol (MCP) defines three cor | |
| MCP Prompts | # MCP Prompts | |
| MCP Resources | # MCP Resources | |
| MCP Tools | # MCP Tools | |
| Mean Time to Restore (MTTR) | MTTR (Mean Time to Restore/Repair) là thời gian tr | |
| memory-hygiene | ||
| Merging Strategies (Agentic) | In agentic systems that use parallel execution, Me | |
| Nhà Minh — Lạm Phát Bạc và Bẫy Tài Chính | Từ giữa thế kỷ 16, bạc từ mỏ Nhật Bản và châu Mỹ đ | |
| Nhà Minh — Sụp Đổ Cơ Sở Thuế Ruộng Đất | Dưới thời Minh, diện tích đất canh tác thực tế liê | |
| Model Context Protocol | # Model Context Protocol | |
| Monolithic Task | A monolithic task is one where a single agent is r | |
| Multi-Agent Topologies | Multi-Agent Topologies describe the patterns of co | |
| Multi-Cloud Architecture | Multi-cloud architecture là chiến lược triển khai | |
| Naive Bayes Classifier | Naive Bayes Classifier (NBC) là một họ thuật toán | |
| Open-ended Goals | Open-ended goals are objectives where the target i | |
| EVPN External Connectivity in OpenStack Neutron | [ | |
| DVR: Distributed Virtual Router in OpenStack Neutron | [“DVR OpenStack”, “distributed routing Neutron”, “ | |
| OpenStack External Network Mapping: Provider Networks to Tenant Overlays | The Two-Network World in Neutron | |
| Floating IP NAT Mechanism in OpenStack Neutron | [“floating IP Neutron”, “DNAT SNAT Neutron”, “one- | |
| OpenStack ML2 Plugin: Cấu hình Overlay Networks | ML2 (Modular Layer 2) là core plugin của Neutron. | |
| OpenStack Neutron Overlay Protocols: GRE, VXLAN, GENEVE | OpenStack Neutron cung cấp **self-service networks | |
| OpenStack Overlay Networks: VXLAN, GRE, GENEVE | archived: true | |
| OpenStack Provider Network Routing (No Floating IP) | Trong môi trường lab hoặc private cloud, floating | |
| OpenStack Trove Guest Agent Connectivity (OVN Context) | Trove guest agent (chạy bên trong database VM) cần | |
| OpenStack Trove PostgreSQL HA Replication | Trove cung cấp managed database service trên OpenS | |
| Orchestrator (Agentic) | In a multi-agent system, an Orchestrator is a cont | |
| Orchestrator Sub-Agent | The Orchestrator Sub-Agent is a workflow pattern w | |
| Orchestrator (Agentic) | An Orchestrator is a controlling agent in a multi- | |
| Over-Decomposition | Over-decomposition is an anti-pattern in agentic d | |
| Over-Engineering | Over-engineering in the context of agentic systems | |
| Over-Escalation Anti-Pattern | Over-Escalation is an anti-pattern in agentic syst | |
| Over-specified | An over-specified handoff message is one that cont | |
| Overfitting và Underfitting | Overfitting và Underfitting là hai thái cực của cù | |
| OVS Bridge Management NIC Pitfall | Một trong những mistakes nguy hiểm nhất khi cấu hì | |
| Packer OpenStack Image Pipeline | Packer với OpenStack plugin cho phép build immutab | |
| Parallel Decomposition | Parallel Decomposition is a task decomposition pat | |
| Parallel Execution (Agentic) | Parallel Execution is a core orchestration pattern | |
| Partial Failure Handling | Partial Failure Handling is a set of strategies us | |
| Patroni | Patroni là một HA agent cho PostgreSQL, chạy trên | |
| Peer-to-Peer (P2P) Topology | The Peer-to-Peer (P2P) topology is a multi-agent o | |
| Perception, Reasoning, Action, and Observation (PRAO) | The Perception, Reasoning, Action, and Observation | |
| Pgpool-II | Pgpool-II là một middleware proxy ngồi giữa applic | |
| Pipeline Topology | The Pipeline Topology is a multi-agent orchestrati | |
| Pirate Metrics (AARRR) | Pirate Metrics là framework của Dave McClure để đo | |
| PostgreSQL HA: Patroni + Pgpool-II Combined Pattern | Khi kết hợp Patroni và Pgpool-II, bạn có một Postg | |
| Progress vs. Spinning | In an agentic loop, progress is marked by new | |
| Prompt Chaining | Prompt Chaining is the simplest workflow pattern, | |
| RAG vs Compiled Knowledge | Hai paradigm khác nhau để LLM làm việc với tài liệ | |
| Reasoning Errors | Reasoning Errors occur when the agentic system gen | |
| Replanning Triggers | Replanning Triggers are events that necessitate a | |
| Replanning | Replanning is the process of revising the current | |
| Resilience vs Robustness | Robustness và Resilience là hai triết lý đối lập t | |
| Result Aggregation | Result Aggregation is a core function of an orches | |
| RHOSO — Red Hat OpenStack Services on OpenShift | RHOSO (Red Hat OpenStack Services on OpenShift) 18 | |
| Role of the Orchestrator | The Role of the Orchestrator is to manage tasks an | |
| Routing (Workflow Pattern) | Routing is a workflow pattern that uses a classifi | |
| Agile Release Train (ART) | Agile Release Train (ART) là đơn vị tổ chức cốt lõ | |
| SAFe Enablers | Enabler trong SAFe là backlog item không deliver b | |
| SAFe IP Iteration | IP Iteration (Innovation and Planning Iteration) l | |
| SAFe Planning Interval (PI) | Planning Interval (PI) — trước đây gọi là Program | |
| ART System Team | System Team là một team đặc biệt trong ART chịu tr | |
| SAFe Value Stream | Value Stream trong SAFe là chuỗi các bước cần thiế | |
| SAFe — Scaled Agile Framework | SAFe (Scaled Agile Framework) là framework phổ biế | |
| Scaling DevOps với SAFe | SAFe cung cấp một bộ khung để s | |
| Schema Versioning | Schema Versioning is the practice of assigning ver | |
| Seekable OCI (SOCI) | Seekable OCI (SOCI) là công nghệ open-source do AW | |
| Semantic Validation | Semantic Validation assesses whether agentic outpu | |
| Separation of Responsibilities (Agentic) | Separation of Responsibilities in a multi-agent sy | |
| Sequential Decomposition | Sequential Decomposition is a task decomposition p | |
| Sequential Execution | Sequential Execution is a workflow where tasks are | |
| Sequential Pipeline | A Sequential Pipeline is an agentic workflow patte | |
| Silent Failure | A Silent Failure is a high-risk error condition in | |
| Site Reliability Engineering (SRE) | SRE là một practice (và job title) xuất phát từ Go | |
| SLSA — Supply-chain Levels for Software Artifacts | SLSA (đọc là “salsa”) là một security framework dạ | |
| Software Architect Role | Software Architect không phải là Developer cấp cao | |
| State Accumulation | State Accumulation is a condition in sequential pi | |
| Static Planning | Static Planning is an approach where all the steps | |
| Stop Signal | A Stop Signal is a form of loop control where a co | |
| Subagent Definition | A sub-agent is a distinct agent instance created b | |
| Swift Single-Node Setup (Kolla-Ansible) | Swift là OpenStack object storage service, mandato | |
| System Design Patterns | Một tập hợp các vấn đề phổ biến trong thiết kế hệ | |
| T-Shaped Developer | A T-shaped developer has deep expertise in one are | |
| Tail Latency | Tail latency là hiện tượng một tỷ lệ nhỏ các reque | |
| Task Assignment | Task Assignment is a core function of an orchestra | |
| Task Completion | Task Completion is a termination condition for an | |
| Task Decomposition | Task decomposition is a foundational technique in | |
| Termination Conditions | Termination Conditions are explicit criteria that | |
| Theory of Constraints (TOC) | Theory of Constraints (TOC) là mô hình quản lý do | |
| Time to Market (TTM) | Time to Market (TTM) đo khoảng thời gian từ khi mộ | |
| Tool Description Routing | # Tool Description Routing | |
| Tool Error Design | # Tool Error Design | |
| Tool Errors | Tool Errors arise from external APIs or services f | |
| Tool Idempotency and Retry | # Tool Idempotency and Retry | |
| Tool Input Schema Contract | # Tool Input Schema Contract | |
| Tool-Use Lifecycle | Tool use is what transforms an LLM from a text gen | |
| Tool Use (Agentic) | Tool use is the mechanism by which an agentic syst | |
| True vs. Artificial Dependency | In agentic workflow design, distinguishing between | |
| Tứ Giác Nước — Mô Hình Đô Thị Sông Nước Việt Nam | Tứ giác nước là mô hình quy hoạch đô thị đặc trưng | |
| Turn Budgets | A Turn Budget is a strict limit on the number of i | |
| Under-specified | An under-specified handoff message is one that lac | |
| Unpredictable Tool Sequences | Unpredictable tool sequences occur when the specif | |
| Validation Gate | Validation Gates are checkpoints within an agentic | |
| Value Stream Mapping (VSM) | Value Stream Mapping là công cụ trực quan từ Lean, | |
| The 4094 VLAN Limit | The IEEE 802.1Q standard defines a **12-bit VLAN I | |
| VXLAN và VNI Scalability | VXLAN (Virtual Extensible LAN) — định nghĩa trong | |
| Well-Defined Subtasks | A well-defined subtask has a single responsibility | |
| Westrum Culture Typology | Ron Westrum là nhà xã hội học nghiên cứu an toàn t | |
| Workflow-First Agentic Architecture | When a task has a predictable input, a known outpu | |
| Workflow Patterns | Anthropic’s “Building Effective Agents” (Dec 2024) | |
| Y Bát (衣鉢) | # Y Bát (衣鉢) | |
| Zero Trust Network in Kubernetes | Zero Trust Network là mô hình bảo mật theo nguyên | |
| Zettelkasten | Zettelkasten (tiếng Đức: “slip box”) là phương phá |