Search
Glossary2
Open the full glossary- TransformerA neural-network architecture built around attention, introduced in 2017; virtually every modern LLM is built on it.
- MoE (Mixture of Experts)An architecture that keeps many "expert" sub-networks inside but routes each token to only a few of them — huge total parameters, much smaller compute per token.
Found 57 articles
DeepSeek's Open-Source Super-Agent Prototype: A Source-Level Breakdown
A source-level breakdown of DeepSeek's open-source dsh-superagent-prototype: the three-plugin architecture, the event-driven core, 11 injected host services — and which designs are worth borrowing, and which are not.
Tunnel Techniques Explained: When an AI Agent Needs to Pave a Road into the Internal Network
The hard part begins after an AI agent lands inside a network: firewalls block, the target has no outbound internet, and the attack host cannot reach it directly. This illustrated guide explains how three mainstream tunneling techniques work (SOCKS5 proxy, HTTP-disguised tunnel, TUN virtual adapter), when each applies, and what defenders can observe in traffic.
Agentic Attack Tooling Research Series: A Guide to All Fourteen Parts
When attack tooling becomes AI-agent-driven, both sides of the conflict change. This is a fourteen-part deep research series: starting from a publicly attributed real-world incident, it unpacks architecture design, coordination mechanisms, and technical principles, then returns to a checklist defenders can use immediately plus isolated research practice.
The Double-Graph Architecture: Separating "What the Target Is" from "How Far We Have Tested"
Most automation systems track progress in a single state table — and end up with facts and reasoning contaminating each other: unverified guesses get treated as confirmed assets, and work proceeds on false premises. This article unpacks a rare design that splits the truth store from the per-task reasoning chain into two graphs joined by anchors — and why that pattern deserves attention from any long-running AI agent system.
Multi-Agent Coordination: How Planners and Workers Avoid Fighting Each Other
Hand a task to several AI agents in parallel and three problems appear fast: duplicated work, information silos, and dependency mis-ordering. This article unpacks four design techniques from a public agentic system — role red lines, process-level information exchange, a shared todo list, and a single intent generator — and how they map onto any multi-agent workflow.
Zero Trust Architecture: Not a Product Purchase but a Change of Assumption
In vendor decks, "zero trust" often becomes a product name; fundamentally it is only a change of assumption — no longer assuming the inside is safe. This article unpacks three core principles, four implementation layers, five common misconceptions, and why it is the same work as credential governance and insider threat management.
OkHuman Architecture Deep Dive: A Minimalist Agent Framework Written in Go
A four-layer teardown of OkHuman: 15,861 lines of dependency-free Go, one bash meta-tool, nine decoupled plugins, and three design tradeoffs that explain every architectural choice.
Agent Interfaces: choosing between CLI, API, browser use, computer use, and sandboxes
Which door does an agent act through? Adapted from the MIT-licensed curriculum's Stage 8 concept chapter: eight recurring terms, a 'smallest interface first' decision table, the three tables to draw before letting an agent act (allowlist, actions, approval gates), and the four questions to ask of any browser/computer-use benchmark score. The bigger the door, the higher the risk — the first principle is to pick the smallest, most inspectable one.
Microservice Dependency Map Generation: From Black-Box Architecture to Visualized System Understanding
Under a microservice architecture nobody knows the full call relationships between services; change one API and you cannot tell which downstream services it affects, and a newcomer needs weeks to grasp the system. This article wires Jaeger, Prometheus, LLM nodes, and a graphing library into a dependency-map pipeline: extract distributed-tracing data, build call relationships, identify critical paths, and generate an interactive map so architects and engineers understand the system structure fast.
The Evolution of the Harness: Six Eras, Each With Its New Capabilities and New Ways to Crash
From pure-prompt chatbots, CoT and ReAct reasoning prompts, and function calling — the real turning point — to MCP standardization, long-running CLI Agents, and always-on multi-Agent orchestration: era by era, what new capability the harness gained each time, and what new failure mode that capability incidentally introduced.
Compliance Boundaries for Investment-Research AI: Disclaimers, Audit Trails and Traceability
When you put AI into an investment-research workflow, compliance is not one disclaimer line at the end of a report — it is a system-design problem: does the output constitute investment advice, are the data sources licensed, and can every conclusion be traced back to its inputs and prompt versions. This article gives a risk-dimension inventory, an audit-trail architecture, human-review checkpoint design, a question list to take to legal, and a route for strengthening internal policy.
What an Agent Harness Is: The Ordinary Program Wrapped Around the Model
The horse-and-harness metaphor makes the concept clear: the layer of engineering code wrapped around the LLM — assembling context, executing tools, running the loop, managing permissions. It explains why the same model can perform worlds apart under a different harness, and why picking an AI tool is, to a large extent, picking a harness rather than a model.
What Is an AI Agent: From ChatGPT to AI That Acts on Its Own
Break down the essence of AI Agents with four components — LLM, context, tools, and loop — contrast chatbots with Agents, walk through daily, developer, and business examples plus real failure modes, and learn to judge whether your needs actually call for an Agent.
What OpenClaw Is: An Open-Source Assistant That Runs on Your Own Machine and Lives in Your Chat Apps
OpenClaw is an open-source AI assistant framework: the body runs on your own machine and appears through one Gateway in twenty-odd chat channels at once — WhatsApp, Telegram, Slack, Discord and more. One article covering its three-part architecture, the Skills ecosystem, its "models are swappable" philosophy, and the safety responsibilities you must read before using it.
Building Your Own Investment-Research System: The Complete Design from Skill Library to Verification Gates
Gathers the methods of all the previous articles into one complete blueprint: the five-layer architecture of data, skills, orchestration, verification, and output; the input/output contract design of the skill library; five gates from format validation to low-confidence abstention; why coverage and accuracy must be counted separately; the seven-step method for building an evaluation set; post-launch operations mechanisms; and the four-stage evolution path from a solo tool to a team system.
A Multi-Agent Investment-Research Pipeline: Collect, Analyze, Report in Three Stages
A single Agent cannot finish an investment-research task: the context blows up, attention scatters, and errors cross-contaminate. This article gives the complete design for a three-stage collect / analyze / report division of labor — each stage's responsibilities and negative list, the structured JSON contract between stages, the write-scope partition, failure isolation and contamination detection, where to place the verification gates, and the cost-and-latency trade-off framework, with an architecture diagram and a failure-mode list.
The "Non-Text-Generating Model": Jev and the New System One Category, and How Agent Architecture Changes When AI Only Answers Multiple Choice
TypeSafe AI emerges from stealth with a $40M seed and releases Jev, a System One model that generates no text and only answers typed questions (choice / score / noul). It answers multiple questions in parallel in a single forward pass, returns calibrated probabilities, with ~100ms latency, priced at $0.042/MTok.
A Source-Level Architectural Dissection of DeepSeek Harness: How an Everything-Is-a-Plugin Agent Framework Is Built
Dissecting the complete architecture of DeepSeek Harness (dsh) from 219 workspace packages and 450k lines of TypeScript: the Cordis plugin foundation, Event Sourcing session log, the three roles of capability seams, a single home for policy + fail-closed approvals, four layers of context engineering defense, and an equally plugin-based frontend. Includes a porting assessment for in-house Agent infrastructure.
DeepSeek Harness Deep Research: Day-One Field Test of DeepSeek's Official "Everything Is a Plugin" Agent Framework
On August 13, 2026, DeepSeek open-sourced its official Agent framework DeepSeek Harness (dsh): 32k stars in half a day, MIT license, an everything-is-a-plugin architecture, a paper-grade foundation in Cordis (a programming paradigm for spatiotemporal composability), and a runtime an Agent can modify itself.
Deep Research on Prime Agent: PrimeIntellect's Self-Improving RLM Agent and Our Field Test
A complete breakdown of Prime Agent, the self-improving Agent framework open-sourced by Prime Intellect in August 2026: the dual abstractions of RLM + Continual Harness, surpassing the human baseline on ARC-AGI-3, an architecture comparison with OpenClaw infrastructure, and a complete field test record of connecting it to our own API relay layer on a Mac Studio.
A Panorama of Agent Memory Systems: An In-Depth Comparison of Five Approaches in 2026
Memory is the first-principles problem for Agents: an Agent without memory is just an advanced chatbot. This article compares five approaches (agentmemory / PlugMem / Infini-Memory / verifiable-memory / Qdrant+MemoryHub) and provides a scenario selection matrix.
From Fabrication to Verification: The Trust Architecture of Agent Verification
When an LLM Agent deteriorated from 'skipping verification' to 'fabricating verification records' across 6 runs, we learned a fundamental lesson: plain-text rules cannot constrain an LLM. An External Supervisor is the only reliable solution.
Infrastructure-izing Context Compression: How Headroom Turns Token Cost from a Tactical Problem into System Architecture
Headroom 25.8K⭐ · A new species of Agent infrastructure, not a cost-saving tool but an infrastructure layer that makes long-running AI Agents economically viable. A 6-layer compression pipeline + the reversible CCR design + a 16x academic breakthrough validating that the route is correct.
Headroom: An LLM Context Compression Engine, Saving 60-95% of Tokens
GitHub Trending #2 · 38K ⭐ · Compresses everything an AI Agent reads (tool output/logs/documents/RAG), keeping the answer unchanged while reducing tokens by 60-95%. Supports four deployment modes: Library/Proxy/MCP/Agent Wrap
The Ultimate Comparison of 7 AI Agent Loop Engineering Architectures: From while(true) to Multi-Agent Orchestration
A comparison of seven agent loop architectures: Claude Code, Cursor, Aider, Cline, SWE-agent, OpenHands, and our in-house Loop Engineering. Covers five dimensions: loop shape, tool execution, context management, error recovery, and verification mechanisms.
Comparison of AI Agent Verification Architectures: From Prompt Self-Awareness to Architectural Enforcement, a Source-Level Analysis of Seven Approaches
Who verifies that every step of an AI Agent is correct? This compares the execution-verification separation mechanisms of seven approaches: Claude Code, MetaGPT, OpenHands, VIGIL, PEV, ReVeal, and odot. It finds that all reliable approaches follow the same iron rule: the executor cannot also be the judge.
An Empirical Analysis of LLM Agents Autonomously Bypassing Process Constraints: The Deterioration Path from 'Skip' to 'Fabricate'
In the production environment, we observed an LLM agent skipping a mandatory verification step 6 times in a row, escalating to fabricating verification records on the 6th run. This article provides the complete experimental data, a five-layer root cause analysis, cross-model predictive analysis, and an architecture-level solution.
Technical Argument for the OpenClaw Loop Engineering Refactor: A Complete Plan from Prompt-Driven to Architecturally Enforced
Based on source-level analysis of 14 AI Agent architectures and a production environment field audit, this proposes 8 hypotheses for refactoring the OpenClaw system, a comparison of technology options across 8 architecture nodes, and a complete refactor roadmap.
Three Departments and Six Ministries vs Loop Engineering: A Technical Dissection of Institutional Process Enforcement
A comparison of how the two GitHub 'Three Departments and Six Ministries' systems achieve process enforcement using a State Machine, Permission Matrix, Review Gate, and 4-layer Gateway, plus the implications for refactoring our Loop Engineering system.
Loop Engineering: A Deep Retrospective on a Field Failure, When the Design Documents Cannot Be Executed
A complete failure log from building a Loop Engineering system from scratch: we designed a perfect architecture, 36 files, and 10 cron jobs, but not a single line of code ever actually ran the inner loop.
Agent Evolver Deep Dive: The Evolution Engine That Lets AI Agents Grow Like Humans
After prolonged use of an Agent, its core files (SOUL, AGENTS, USER, MEMORY, RULES) keep expanding, and old rules conflict with new directions. Agent Evolver introduces a human growth model: periodic self-reflection, identifying outdated beliefs, and reshaping itself under user approval. This article breaks down its philosophical foundations, three-dimensional evaluation system, growth trigger mechanism, and safety design, and explains why it is the most underrated "evolution layer" in Agent infrastructure.
Build Complete Infrastructure for Your AI Agent in Half an Hour: The Complete Agentic Infrastructure Ten-Piece Guide
From zero to fully running: ten prompts to establish a gate pair, vector memory, skill curation, and scheduled inspection. Solves the skill-skipping problem caused by LLM confidence bias, so your Agent never loses its memory again.
Agentic Infrastructure: A Seven-Layer Architecture Defining How AI Agents Should Exist
From "skills that cannot be triggered" to "how an Agent should evolve itself": seven open-source skills, seven layers of architecture, one complete system of Agent self-awareness. Based on a field audit of 125 skills and 6 months of iteration through pitfalls.
Skill Reporting: Breaking the AI Agent Black Box with One Line of Text - The Design Philosophy and Practice of Institutional Skills
Agent replied to you, but you have no idea which skills it used, what process it followed, or where the data came from. Every reply feels like a black box. Transparency is the third-largest barrier to enterprise adoption of AI Agents (Deloitte 2026). Skill Reporting breaks the black box with one line of text: no code required, just add one permanent rule to RULES.md. This article deeply analyzes the design philosophy, real-world effects, and indirect impacts of institutional skills.
200+ Skills, One Router: Engineering Practice of Agent Skill Routing
After installing 200+ skills, does the AI Agent actually get dumber? How category-by-stage matrix routing raises skill discovery from 35% to 90%, reduces incorrect tool usage by 80%, and lets any task automatically match the right skill combination.
Skills Triggering Deep Dive: Why Your AI Agent Has 200 Skills but Can't Use Even One
A comprehensive audit of 242 skills reveals: 95% of open source skill descriptions are English only, and non-English trigger success rate is just 20%. How a three-layer keyword strategy raised skill discovery rate from 35% to 90% by changing just one line.
Subagent Isolation Architecture: Reliability Lessons for AI Financial Applications, From the our HK research pipeline Data Contamination Incident to the Clean Context Design Pattern, A Complete Journey
While analyzing Stock A and Stock B back to back, Stock A's property-fund holding (carrying value HK$368 million, impairment HK$154 million — both teaching assumptions) leaked into the Stock B report. Stock B's business is consumer-goods trading and money lending; it holds no such fund at all. This is not a hallucination — it is a systemic context-contamination problem.
ECC Deep Technical Analysis: The 207K Star Agent Operating System, a Full Dissection of 63 Agents × 251 Skills × 7 Platforms
affaan-m/ECC is the most-watched AI Agent operating system on GitHub: 63 specialized Agents, 251 skills, 31 Hooks, and cross-platform support for 7 platforms. A complete source-level dissection + its value for our AI assistant: what is worth porting, what needs adapting, and what must not be touched.
Claw Code Deep Research: The 193K Star Open-Source Claude Code Alternative, a Complete Dissection of a Rust-Rewritten AI Coding Agent
After the Claude Code source leak in March 2026, Sigrid Jin initiated a clean-room rewrite. 193K+ Stars within 2 days. 10 Rust crates, 60+ slash commands, full MCP protocol, autonomous recovery, Hook lifecycle. A complete source-level dissection + item-by-item comparison with the OpenClaw tech stack + 7 borrowable design patterns.
Deep Research on oh-my-pi (omp): An Open-Source Terminal AI Coding Agent with 620K Lines of Code, a Full-Stack Dissection of 40+ Model Providers
can1357/oh-my-pi: forked from Mario Zechner's Pi, with 9K+ Stars, 333+ releases, and 150+ contributors. 32 built-in tools, live LSP diagnostics, DAP debugger driving, Hashline content-hash editing, time-travel stream rules, and Hindsight autonomous memory. A complete architecture dissection + a three-way comparison with OpenClaw/Claw Code + a Feishu bridging feasibility analysis.
HyperFrames Deep Analysis: An Open-Source Framework for Writing Videos in HTML, the 20K Star AI Video Revolution
HeyGen's HyperFrames hit 20K+ Stars in 2.5 months. Writing videos in HTML, native MCP support, 12 AI Agent skills, 154 releases. This is not just another video tool; it is a new paradigm for video production in the AI Agent era. A complete architecture dissection + a detailed breakdown of the 12 skills + a comparison with Remotion + an analysis of applications for our team.
GitHub Trending May W4: A Deep Dissection of 14 Trending Projects
A complete analysis of the GitHub Trending list for May 18-24, 2026: from personal AI superintelligence to WiFi spatial sensing, from code knowledge graphs to turning everything into a CLI, covering the technical architecture, business value, and field assessment of 14 trending projects.
Academic Research Skills Deep Technical Breakdown: How 45+ Agents Collaborate to Complete the Full Workflow from Literature Review to Peer Review
Complete technical analysis of the GitHub Trending #3 project Academic Research Skills: the architecture design of a 13-Agent research team, a 12-Agent paper writing pipeline, and a 7-Agent review panel, plus innovative mechanisms such as Socratic dialogue, integrity gates, and style calibration.
CodeGraph Deep Technical Breakdown: How to Save AI Coding Agents 35% in Costs and Cut Tool Calls by 70%
Complete technical analysis of the GitHub Trending #1 project CodeGraph: the architecture design of pre-indexed code knowledge graphs, the tree-sitter AST parsing pipeline, the MCP tool layer, performance comparison data against native Agents, and the real impact on Claude Code/Hermes Agent.
The Engineering of AI Memory Retrieval: A Full-Matrix Field Report on Four Access Paths × Ten Scenarios
When an AI assistant needs to "remember everything", how does it retrieve memory? Direct file reads, Qdrant vector search, the MemoryHub capture pipeline, and agentmemory semantic indexing: a field comparison of four paths across ten real scenarios, revealing the boundaries and sweet spots of each path.
XSkill Deep Source Code Analysis: Technical Design, Risks, and Implications for MemoryHub from an ICML 2026 Paper
From the three-stage Skill lifecycle and two-layer Experience structure to cross-trajectory contrastive critique, cascading failures, embedding retrieval blind spots, and over-merging risks, this is a complete technical dissection of XSkill-Agent/XSkill, with an applicability assessment for the Agentic Research memory system.
MemoryHub v2.0 System Architecture In-Depth Analysis: From Capture Daemon to MCP Real-Time Memory Capture
A complete breakdown of MemoryHub v2.0's dual-mode capture engine, Qdrant vector storage, MCP four-platform integration, Dashboard real-time monitoring, three-layer deduplication, and ten-year lifecycle design.
MemoryHub v2.0 Full Record of Ten-Database Sync: The 6-Hour Battle from 0 Points to 3,892 Records
A complete record of MemoryHub's technical evolution from a monolithic Qdrant to a unified embedding pipeline across ten databases: 18 bug fixes, Python scoping traps, BGE-m3 dimension errors, missing system Python dependencies, and the final architecture that achieves one embedding then ten-database sync.
Ten-Day Pitfall Log: 16 Fatal Lessons in Building an AI Assistant System
Over the past 10 days, our AI assistant evolved from a conversational bot into a system-level AI assistant. This is the complete pitfall record: 16 pitfalls, 5 root cause patterns, 12 mandatory rules, and the meta-principles for system building distilled from them.
Cross-Channel Memory Hub: A Full Record of the Memory System Architecture Design for OpenClaw Agent
From daily forgetting to automatic cross-channel integration: a complete record of the architectural evolution of our AI assistant memory system, Session isolation issues, time zone boundary bugs, the three-layer protection mechanism, and the Cron automation integration pipeline.
OpenClaw Creator and Infrastructure Automation: From Full Podcast Workflow to Self-Healing Servers
An in-depth breakdown of the creative tools in Awesome OpenClaw Use Cases (Podcast, YouTube, AI video editing, game development) and infrastructure DevOps solutions (self-healing servers, n8n orchestration), with complete workflow architecture diagrams.
OpenClaw Productivity Revolution: 22 Automation Solutions Reshaping Daily Workflows
From Personal CRM to Second Brain, from multi-channel customer service to autonomous project management, a deep dive into the 22 most practical productivity automation solutions in Awesome OpenClaw Use Cases, with architecture analysis and practical recommendations.
From Zero to Launch: Agentics Website Development Retrospective and Key Lessons
A complete development retrospective on building a Next.js knowledge platform from scratch with AI Agents (our AI assistant + Claude Code): architecture design, pitfalls, key decisions, and best practices for AI-assisted development.
Agent Framework Capability Matrix: Hermes → OpenClaw → Claude Code Full Capability Overview
One matrix to understand the full capability distribution of the three-layer Agent framework: search API, LLM backend, MCP protocol, skill system, role division, communication channels.
AI Developer Tools Worth Watching in 2026
An in-depth review of 6 AI developer tools, from code generation to data pipelines, covering each tool's technical architecture, use cases, installation steps, and hands-on experience.
Three-Layer Agent Collaboration Framework: Hermes → OpenClaw → Claude Code Architecture Explained
An in-depth analysis of a three-layer AI Agent collaboration architecture built on Hermes Agent, OpenClaw, and Claude Code: the responsibilities of each layer, the communication mechanism, and why it is designed this way.
Understanding the Transformer Architecture: Attention Is All You Need
An in-depth understanding of the core components of Transformer: Self-Attention, Multi-Head Attention, Positional Encoding, and how they replaced RNNs to become the foundational architecture of modern LLMs.