Personal agent runtime with tools, memory caches, skills, cron, Telegram, email, browser, repo, and vault access.
Tech Stack / AI Agents
AI Agents
How Hermes, coding agents, reusable skills, and command routing fit into Viggo Meesters’ AI-assisted workflow stack.
Current shape
Deep coding and repo implementation lanes, used when the work needs sustained edit/test loops.
Procedural memory for recurring task types, with routing, constraints, pitfalls, and verification gates.
Why this belongs in the stack
Agents are the execution layer for coding, research, automation, review, and command routing.
It matters when the default agent runtime, coding agent, or routing layer changes.
What stays out
Skills are procedural memory and belong in the Skills registry.
Capture routes, schemas, and validation belong under Personal Knowledge System.
How it earns a place
Agent surfaces belong here when it changes delegation, verification, code execution, research, or automation routing; keep model experiments out unless they become operational.
The page is accurate when each agent has a clear job boundary and does not blur into Personal Knowledge System components or implementation details.
It matters when the active agent runtime, delegation model, voice/Telegram routing, or coding assistant changes the way tasks are executed and verified.
Concrete example
Hermes is listed as an agent runtime and command bus, while individual implementation details are represented through the Skills registry and not copied into this stack page. The boundary is intentionally public-facing: agent surfaces are named by role, not by private history or unpublished implementation details.