protologue

Coding Agent

Also called software engineering agent, agentic coding, AI coding assistant.

A coding agent is an AI agent that works in a software repository by reading files, editing code, running commands and tests, and iterating on the results to complete programming tasks.

Description

Research found that the interface matters as much as the model. Tools designed for a model, with concise file viewers, search, and editing commands that report errors, improved results over a raw shell. Coding agents are commonly evaluated on SWE-bench, which asks them to resolve real GitHub issues so that hidden tests pass.

Sources

  1. Yang et al. (2024). SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering.
  2. Jimenez et al. (2023). SWE-bench: Can Language Models Resolve Real-World GitHub Issues?.

Cite this entry

Protologue. (2026). Coding Agent. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0145). https://protologue.com/t/coding-agent/

BibTeX
@misc{protologue_coding_agent,
  title = {Coding Agent},
  author = {{Protologue}},
  year = {2026},
  howpublished = {Protologue: A Taxonomy of Prompting and LLM Techniques, v1.4.0},
  note = {Entry PTL-0145},
  url = {https://protologue.com/t/coding-agent/}
}

Markdown JSON