The File Changed. The Agent Didn’t: Inside Hermes Prompt Assembly
Hermes Agent Architecture, Part 2: a reproducible teardown of context files, prompt snapshots, and compression.
The Agent Stack
The Agent Stack is a technical deep dive into how modern AI agents actually work under the hood, from control loops and memory systems to orchestration, evaluation, and production deployment.
Live Last read · last published · next check
Latest posts
Hermes Agent Architecture - Part 1: Gateway, Sessions, and the Agent Loop
How one message becomes a stateful run across CLI, chat, and remote execution
The Data Agent Stack - Part 7: Evals, Provenance, and Production Feedback Loops
Why a data agent is only useful if you can prove where the answer came from
The Data Agent Stack - Part 6: Trust, Permissions, and Governance
Why data agents inherit every governance problem you already had
The Data Agent Stack - Part 5: Tools, Query Execution, and the Analyst Loop
Why fewer tools beat more tools
The Data Agent Stack - Part 4: Meaning Lives in Code
Why schemas describe shape, but pipeline code explains meaning
The Data Agent Stack - Part 3: Context Assembly for Data Agents
Why the hard part is not the model call, it’s what you put in front of it
The Data Agent Stack - Part 2: The Data Foundation Is the Agent
Why no agent architecture can save messy data
The Data Agent Stack - Part 1: What Is a Data Agent?
Why a data agent is not just text-to-SQL
The Agent Stack - Part 8: Observability, Evaluation, and Production Feedback Loops
Why demos are stories, but production needs evidence