Large language model (LLM) agents are increasingly operating through terminals, yet existing surveys scatter terminal‑mediated behavior across software engineering, tool use, and computer‑use research. We define terminal agents as systems whose dominant observation‑action loop is mediated by terminal command execution, textual feedback, and stateful environment interaction. Using terminal‑mediated execution as an organizing lens, this survey delineates workload‑level boundaries and connects system architecture, competence acquisition, and evaluation via a seven‑dimensional terminal competence profile.
Our synthesis shows that realized behavior is jointly shaped by the model, interface, harness, runtime, and environment. Executable trajectories ground learning in action consequences, verification, and recovery, whereas prevailing evaluations emphasize final outcomes, exposing uneven evidence for process quality, recovery, and governance. Bounded fixed‑condition diagnostics illustrate two implications: (1) benchmark families reveal different process signals; (2) matched system comparisons expose benchmark‑dependent performance and limits of component attribution.
These findings motivate explicit reporting of system and runtime conditions, supported by replayable traces and process‑level evidence. The framework offers a unified basis for studying terminal‑mediated agency across software engineering and emerging application domains.