The model is not the agent — eight runtime components plus a feedback loop make the harness

Asked:

“What are the components of agent harness and how do they contribute to the harness?”

A synthesis of 10 component groups drawn from live web sources (2026-09-02): a source-code study of 11 coding-agent systems cross-checked against official OpenAI, Anthropic, LangGraph, Google ADK, and Microsoft Agent Framework materials. Eight groups form the core runtime and control surfaces; observability and evaluation close the production feedback loop.

How the ten components fit together

core runtime flow state feeding the loop control & coordination production feedback loop hover a component for its contribution and sources · badge = independent pages citing it
The harness, not the model, does most of the operational work: a source-code study of 11 coding-agent systems anchors 7 of the 10 component groups. arxiv.org
Safety, permissions and human oversight, together with orchestration, are the most broadly corroborated groups — 4 independent pages each. openai.github.io
Persistence is what lets long-running agents pause for people or external events and resume after failure without starting over. docs.langchain.com
Observability and evaluation are production additions rather than core anatomy: traces and replayable eval runs feed regressions back into the design of the other eight. developers.openai.com

The ten component groups

#LayerComponent & contributionTerminology in sourcesPagesSource
1FoundationModel integration & instructions. Connects the harness to one or more model providers, assembles instructions and prompts, and handles routing, caching, structured responses, and model-specific protocols.Augmented LLM, LLM integration, model clients3arxiv.org
2RuntimeAgent loop & runner. Runs the perceive-reason-act cycle: call the model, interpret its decision, execute actions, feed results back, and enforce stop, timeout, cancellation, retry, and failure-recovery conditions.Agent loop, Event Loop, Runner3arxiv.org
3CapabilitiesTools, actions & environment. Defines what the agent can do beyond text—such as search, code execution, file editing, database/API calls, and interaction with external systems—and returns action results to the loop.MCP clients, Tools & actions2arxiv.org
4StateMemory, context & retrieval. Builds the model context for each turn, manages the finite context window, retrieves relevant knowledge, and preserves short-term conversation state or long-term knowledge across sessions.Comprehensive memory, Memory & context, agent session, context providers3arxiv.org
5StatePersistence & durability. Checkpoints execution and state so long-running work can survive failures, pause for external events or people, and resume without starting over.DBOS, Dapr, Persistence, Restate, Resume Agents, Temporal3docs.langchain.com
6ControlSafety, permissions & human oversight. Constrains risk by validating inputs, outputs, and tool calls; applying least-privilege permissions and sandboxes; setting budgets; and escalating sensitive actions for human approval.Guardrails, Human-in-the-loop, Input guardrails, Output guardrails, Safety & permissions, Tool guardrails4arxiv.org
7CoordinationWorkflow & multi-agent orchestration. Decomposes work, routes tasks, coordinates parallel or sequential steps, delegates to specialized agents, aggregates results, and manages handoffs.Agents as tools / Handoffs, Orchestration, Orchestrator-workers, Parallelization, Routing, Workflows4arxiv.org
8IntegrationExtensibility & integration surface. Lets developers add providers, tools, skills, hooks, middleware, plugins, and standard protocol connections without redesigning the core runtime.Extensibility, MCP clients, middleware2arxiv.org
9OperationsObservability & tracing. Records end-to-end runs and timed spans across model calls, tools, guardrails, and handoffs so teams can debug behavior, audit decisions, and monitor latency, cost, and failures.Spans, Traces, traces2developers.openai.com
10QualityEvaluation & feedback. Uses graders, datasets, replayable eval runs, and feedback loops to measure task success, detect regressions and failure modes, and compare changes before deployment.Evaluator-optimizer, datasets, eval runs, graders2developers.openai.com

Method: 10 normalized component groups synthesized from live web research on 2026-09-02, cross-checked against a source-code study of 11 coding-agent systems and official OpenAI, Anthropic, LangGraph, Google ADK, and Microsoft Agent Framework documentation. "Pages" counts independent web pages supporting each group (2–4 per row). Frameworks use differing labels, so the groups are a synthesis, not a claim of identical terminology; full supporting URL lists were trimmed to one primary link per row for space.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.

Keenable SELECTAsk your own question
Made with Keenable SELECT