Single-agent baseline that gives one Codex CLI session the complete search loop inside the task's own image: Codex edits and evaluates, while Galapagos independently scores every observed checkpoint.
name: codex
display_name: Codex
organization: "default"
type: test_time_search
family: single_agent_baseline
summary: "Single-agent baseline that gives one Codex CLI session the complete search loop inside the task's own image: Codex edits and evaluates, while Galapagos independently scores every observed checkpoint."
source: "Galapagos synthesis of its Claude Code controller and Harbor's installed Codex agent: codex exec --json runs beside a root-owned trusted evaluator in the task container"
tags: [baseline, single-agent, coding-agent, codex, openai, cli, docker, sandboxed]
license: Apache-2.0
controller: galapagos.scaffolds.codex.scaffold.CodexScaffold
components:
population: {kind: keep_all}
selection_policy: {kind: agent_delegated}
prompt_builder: {kind: one_shot_task_brief}
proposer: {kind: codex_cli}
evaluator: {kind: task}
memory: {kind: none}
requirements:
cli: "@openai/codex (installed into Docker task images automatically; required on PATH for local evaluation mode)"
auth: "ChatGPT-managed only: CODEX_ACCESS_TOKEN first (Business/Enterprise), otherwise a refreshable file-backed `codex login` session at ~/.codex/auth.json; API-key billing is not supported"