EB-2
The harness is the body around the model's brain — the orchestration loop, the tool layer, state, permissions, and every configuration surface that turns a chat model into an autonomous engineering agent. In current practice the harness has become the critical infrastructure layer, and the same model can perform very differently depending on the harness it runs in.
Theme: Orchestration →16 topics — scope and references for each.
The harness is everything wrapping the model. Students learn the core agent loop (the model proposes an action, the harness executes it and feeds back the re…
Two fundamentally different modes of running a harness. In **interactive** mode the human types a prompt, reviews the output, and continues the loop — stayin…
Empirical evidence that the harness matters as much as the model. Benchmarks that hold the model constant and vary the harness show large differences in auto…
A project-level instruction file that persists across every session, letting a team encode architecture, conventions, preferences, and forbidden patterns onc…
Skills package reusable behaviours as a folder containing a `SKILL.md` file plus optional supporting scripts and templates. Frontmatter controls invocation:…
Explicit, repeatable entry points for common workflows. Students learn to turn repeated prompts into named commands and to parameterise them with argument su…
Hooks attach deterministic code to specific moments in the agent's lifecycle, so a rule enforces itself rather than relying on the model to remember it. Stud…
Sub-agents are specialised workers that run in their own context window; their verbose output (file searches, log dumps, intermediate reasoning) stays isolat…
The Model Context Protocol is the emerging standard for connecting agents to external tools and data sources. Students learn the client/server model, the dis…
A clarifying conceptual hierarchy, taught through the canonical open-source example. A **framework** provides abstractions for building agents; a **runtime**…
A survey of the major terminal/CLI harnesses tied to a frontier model and a managed harness. Students compare their architectures and sweet spots — deep reas…
The provider-agnostic, bring-your-own-key alternative. Students learn the open-source CLI options — including the de facto open-source standout for terminal-…
The distinction between harnesses you *use* and harnesses you *build on*. LangGraph is a production-grade runtime providing durable execution, streaming, and…
Harnesses that live inside the editor. Students learn the tradeoffs of IDE integration (familiar surface, low ramp, one-button agent mode) versus terminal ag…
A strategic decision rather than a tool choice. Some harnesses lock to a single model provider; others let you bring your own key and route across providers.…
The capstone of the module: composing the surfaces into a coherent configuration for a real project, and treating the configuration directory as shared team…
Topics are combined ad hoc for corporate programs and workshops — there is no fixed public duration. Tell us your goals and we design the track.
Discuss a custom program