Harness Architecture: Turning Autonomous AI Agent Experiments into Reliable Systems
A source audit of revfactory/harness, evaluating multi-agent team topology, role boundary enforcement, dynamic skill generation, and evaluation gates.
Fodev JEO
LATEST INVESTIGATIONS
Source audits & field notes
Codex Plugins as Runtime Bundles: Manifest Contracts and MCP Tool Execution
A source audit of OpenAI Codex plugin architecture, analyzing manifest schemas, runtime capability bundles, MCP tool contracts, and agent execution boundaries.
Nine Essential Claude Code GitHub Repositories for Agent Harness Engineering in 2026
A source audit of nine essential GitHub repositories for Claude Code in 2026, evaluating persistent memory, n8n automation, LightRAG, and orchestration harnesses.
NemoClaw’s Personal Tier Is Broader Than Web Access
NVIDIA’s Personal tier allows any sandbox binary TCP access on ports 80/443, subject to exclusions. Here’s the boundary, remaining controls, and one-host preview scope.
Designing AI Agent Teams with Codex Subagents: A Practical Guide to Multi-Agent Workflows
A technical source audit of OpenAI Codex Subagents, analyzing TOML-based agent definitions, sandbox permission isolation, concurrency limits, and team workflows.
OpenDev and the Case for Compound AI Agents: Multi-Model Architecture in Rust
An architectural source audit of OpenDev, analyzing its Rust-native compound agent design, multi-model workflow routing, and independent critique verifiers.
Deploying NanoClaw in Docker Sandboxes: A Practical Security-First Guide
A step-by-step guide to run each NanoClaw agent in isolated Docker Sandboxes, with hardening checks and real operational patterns for small teams and production rollouts.
OpenClaw Skills Architecture: Managing Frontend Plugins at Catalog Scale
A source audit of the OpenClaw skill ecosystem, evaluating catalog curation, ClawHub package management, permission boundaries, and frontend team adoption.
pi as the Core: Building Hackable Coding Agent Harnesses with TypeScript Extensions
A technical source audit of Mario Zechner’s pi agent and Armin Ronacher’s analysis, examining minimal micro-kernels, lifecycle hooks, and TypeScript extensions.
Fabric Architecture: Modular Prompt Operations and Unix Pipeline Integration
A source audit of Daniel Miessler’s Fabric, evaluating prompt-pattern modularity, Unix stdin pipe composability, vendor-agnostic routing, and content operations.
인기 태그
Search article titles, categories, and tags. Full text is searched when needed.
Type to search articles.