Answer: You validate Gemini CLI behavioral evaluations by running `npm run eval:validate`, a lint-like checker that validates eval source files against structural guidelines and best practices. Errors (file-naming, valid-policy, suite-metadata, prompt-presence, case-name-static, invalid-tool-refs, positive-assertion, workspace-setup) block CI builds and return exit status 1, while the Warning (new-evals-policy) is logged with ⚠ and does not cause the CLI process to exit with status 1. Status: Confirmed Entry: tool/gemini-cli/commands-and-shortcuts/validate-evals-rules Revision: 1 Digest: sha256:ee3b235cd5aa134d (changes whenever the content changes) Timeline: Created: 2026-11-13T03:17:15Z Last changed: 2026-11-13T03:17:15Z Last verified: 2026-11-20T12:50:28Z Next check: 2026-11-21T12:50:28Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of validation command | npm run eval:validate | [1] | 2026-11-20 custom scope command | npm run eval:validate -- evals/my-test.eval.ts | [1] | 2026-11-20 error severity behavior | Errors (`✗`) will block CI builds and return exit status `1`. | [1] | 2026-11-20 warning rule | new-evals-policy | [1] | 2026-11-20 valid policy values | ALWAYS_PASSES`, `USUALLY_PASSES`, or `USUALLY_FAILS | [1] | 2026-11-20 file naming rule | *.eval.ts | [1] | 2026-11-20 Sources: [1] Google. "Behavioral Evaluations & EDK Guide". https://geminicli.com/docs/behavioral-evals.md Role: primary. Accessed: 2026-11-20T12:50:28Z Quote: "A lint-like checker that validates eval source files against standard structural guidelines and best practices. - **Usage:** ```bash npm run eval:validate" (Supports: validation command) Quote: "npm run eval:validate -- evals/my-test.eval.ts" (Supports: custom scope command) Quote: "Errors (`✗`) will block CI builds and return exit status `1`." (Supports: error severity behavior) Quote: "| `new-evals-policy` | **Warning** | New evals must not use `ALWAYS_PASSES` policy initially" (Supports: warning rule) Quote: "Policy must be one of `ALWAYS_PASSES`, `USUALLY_PASSES`, or `USUALLY_FAILS`." (Supports: valid policy values) Quote: "File must match `*.eval.ts` or `*.eval.tsx` naming conventions." (Supports: file naming rule) Published: 2026-10-16T02:36:07Z Snapshot SHA-256: 2124cd48c1c85f7a5401a6ba1094ccd5741dc6eedcdadc069b6e7c3a39a0e49e (captured 2026-11-20T12:50:27Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-13T03:17:15Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T12:50:28Z. No independent second source is configured for this entry Last successful check: 2026-11-20T12:50:28Z Last attempt: 2026-11-20T12:50:28Z Notes and limitations: Extracted automatically from a single official source (section "2. `npm run eval:validate`"). Related: https://latestn.com/tool/gemini-cli/commands-and-shortcuts/vi-mode-shortcuts How do I enable or use Vi mode shortcuts in Gemini CLI? https://latestn.com/tool/gemini-cli/commands-and-shortcuts/view-cached-token-savings How can I view cached token savings in Gemini CLI? https://latestn.com/tool/gemini-cli/commands-and-shortcuts/view-todo-list-shortcut How do I view the full todo list in Gemini CLI? https://latestn.com/tool/gemini-cli/commands-and-shortcuts/run-shell-command-arguments What arguments does the Gemini CLI run_shell_command tool accept? https://latestn.com/tool/gemini-cli/commands-and-shortcuts/start-acp-mode How do I start Gemini CLI in ACP mode? https://latestn.com/tool/gemini-cli/commands-and-shortcuts/start-complex-task-plan-mode How do I start a complex task in Plan Mode in Gemini CLI? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-13T03:17:15Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules POST: https://latestn.com/dispute/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). How do I validate Gemini CLI behavioral evaluations and what are the validation rules and severities? Revision 1, last verified 2026-11-20T12:50:28Z. https://latestn.com/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules (accessed 2026-11-21T10:01:55Z). @misc{LatestN_tool_gemini_cli_commands_and_shortcuts_validate_evals_rules_r1, title = {How do I validate Gemini CLI behavioral evaluations and what are the validation rules and severities?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules}}, note = {Revision 1; digest sha256:ee3b235cd5aa134d; last verified 2026-11-20T12:50:28Z; accessed 2026-11-21T10:01:55Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/gemini-cli/commands-and-shortcuts/validate-evals-rules