Archive
All notes
Every article in the lab. Filter by category, or search by title, summary, or tag.
- Workflows
Build a Cron-to-Approval Loop with Hermes Agent
A practical recipe for running scheduled Hermes Agent jobs that produce useful artifacts, deliver them for review, and wait for explicit human approval before publication or deployment.
- Tools
AWS Bedrock Agents vs LangGraph: A Practical Decision Guide
A builder-focused comparison for choosing between AWS Bedrock Agents and LangGraph when designing tool-using, approval-aware agent workflows.
- Production
Design Agent Memory Without Building a Junk Drawer
A practical guide to designing agent memory: what to store, what to forget, how to retrieve context, and how to keep memory auditable in production systems.
- Workflows
OpenCode Repo Checklist: Safer Coding-Agent Runs
A practical checklist for using OpenCode inside real repositories: constrain context, keep diffs small, review tool use, and verify changes before merge.
- MCP
MCP Resources vs Tools: Keep Read Paths Separate from Action Paths
A practical foundation for designing MCP integrations by separating read-only resources from action-oriented tools.
- Production
Evaluate Agents with Task Traces, Not Vibes
A practical foundation for evaluating agentic systems by capturing task traces, scoring behavior, and turning regressions into release gates.
- Tools
GitHub Connector Checklist for Code-Review Agents
A practical checklist for wiring a code-review agent to GitHub with scoped permissions, webhook handling, review comments, check runs, and approval gates.
- Workflows
LangGraph Checkpoints for Human Approval Gates
A practical recipe for adding approval gates to LangGraph agents with checkpoints, interrupts, and resumable tool execution.
- Foundations
Claude Code vs Codex vs Hermes Agent
Claude Code, Codex, and Hermes Agent all run agentic workflows, but they optimize for different jobs.
- MCP
Tools, Skills, MCPs, and Connectors Explained
Tools let agents act. Skills teach agents procedures. MCP standardizes tool/context access. Connectors link agents to apps and APIs.
- Foundations
What Are Agentic Systems?
Agentic systems are AI systems that can use tools, follow goals, check results, and work across multiple steps.
- MCP
What Is MCP?
MCP is a standard protocol that lets agents connect to tools, data, and prompts through servers.