Setting Up the TWG AI Benchmark CLI

video1.0<iframe src="https://www.loom.com/embed/9e6fb09c9be94122afbc599c8442b3e7" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/os/loom-media-content/loom/a436116f-02ce-4520-8fbb-7301462a1674/1621dc1f-3702-4525-84ba-fad829566798/sessions-thumbnails-9e6fb09c9be94122afbc599c8442b3e7-1f64183d6ad4d7b3.gif/binary?authToken=eyJraWQiOiJtaWNyb3MtZ3JvdXAvdGRwLW9zLWFyY2hldHlwZS85MHZldmJxa2E0ODZhM3QxIiwiYWxnIjoiUlMyNTYifQ.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.WmJDJDHBXOxWrl9N8Y46f_Z2CzELaT09l87zodZ6rFi3Srv7WweIrrV27OQJAP4lpzQqNc9C1YN4eVs296J4Ur_2U92G3fKzqYDoKssHIMZYNrj7lK0XJpogx_CmzJSPyk8Pfi3dadYu2Xb7BdRHA8TxhSquQKsKVvyGkXyAOgt_tV2z7l8QPC-UM4oUDQuinD-yYh_b2ALoIbIvZaD-EtPHr1x5F4JP8KhvOJaBS1tmneX6uJ7tnmuK5I2rcgakkdGXuLuBFWefYoeK9MrqSfHVb2hexFC73OFg984yp_kgNlM2nBh8WZMuzvSliX9wUPkEU4k3iUf3g5d7rfOphw&Expires=1785857855&Signature=ihrYzdpv4m7phLQlv8flbjKdQfnnXENjdsqK%7E8mWYjLsebeERWFN3Aod5lG-zIeoEUnZQHOypimWDxUsn3uGB29Yl1KnkfVhiXMQUBsaIqO6jBsm24KyksQM9Hp-UYGMUydm89hcpwdshSikPFjKkcyYwRTNmv-b22a1PMAugOdZh81eCIMJIpJy5s8hdCmNQywCOlxNvHxml0stFqwjJaQVSzklEDukZ418-JTLZXkjlhSeK2XWF7UwHMLiCsG2AzV079qYmNXKa%7EPerz0o56JLmUdTtz6ADWef25yh86a8bj3l10tpmuhdnwiUetnts2WUnL2rLk0XKo8Ee7nG3A__&Key-Pair-Id=KQOSYIR44AIC0510.872Setting Up the TWG AI Benchmark CLIThis Loom explains how to set up and run a benchmarking CLI tool in an Agent Codex workflow. The speaker verifies the tool is on the latest version, notes that updating requires a benchmark update, and then checks configuration status for auth, prompt inputs, and connectors. In the setup, they choose default pre-save prompt inputs, connect only GitHub by creating a fine-grained personal access token via a browser pop-up, and decline all other connectors. They confirm the setup is ready using the status command showing pair ready yes, then run a benchmark using the TWG AI context benchmark skill with a custom prompt asking what to work on today.