Answer: You configure approval policies and sandbox modes by setting `approval_policy` (options include "on-request", "never", or a granular object) and `sandbox_mode` (e.g., "workspace-write") in the config, optionally with `approvals_reviewer` set to "user" or "auto_review". Status: Confirmed Entry: tool/codex/permissions-and-security/approval-policies-sandbox-modes Revision: 3 Digest: sha256:c5c4102d3ba233c9 (changes whenever the content changes) Timeline: Created: 2026-11-13T03:48:19Z Last changed: 2026-11-17T13:59:32Z (auto-review setting: — → auto_review) Last verified: 2026-11-20T13:35:58Z Next check: 2026-11-21T13:35:58Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of approval_policy | on-request | [1] | 2026-11-20 approvals_reviewer | user | [1] | 2026-11-20 sandbox_mode | workspace-write | [1] | 2026-11-20 allow_login_shell | false | [1] | 2026-11-20 granular approval policy | approval_policy = { granular = { ... } } | [1] | 2026-11-20 auto-review setting | auto_review | [1] | 2026-11-20 Sources: [1] OpenAI. "Advanced Configuration". https://learn.chatgpt.com/docs/config-file/config-advanced.md Role: primary. Accessed: 2026-11-20T13:35:58Z Quote: "approval_policy = "on-request" # Other options: never or { granular = { ... } }" (Supports: approval_policy) Quote: "approvals_reviewer = "user" # Or "auto_review" for Auto-review" (Supports: approvals_reviewer) Quote: "sandbox_mode = "workspace-write"" (Supports: sandbox_mode) Quote: "allow_login_shell = false # Optional hardening: disallow login shells for shell tools" (Supports: allow_login_shell) Quote: "You can also use a granular approval policy (`approval_policy = { granular = { ... } }`) to allow or auto-reject individual prompt categories." (Supports: granular approval policy) Quote: "Set `approvals_reviewer = "auto_review"` to route eligible interactive approval" (Supports: auto-review setting) Snapshot SHA-256: 509cb889c8d02037082a87c240864f0e2618d38ab224c28b562712b2744fbd65 (captured 2026-11-17T13:59:29Z) 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 2026-11-17T13:59:32Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T13:35:58Z. No independent second source is configured for this entry Last successful check: 2026-11-20T13:35:58Z Last attempt: 2026-11-20T13:35:58Z Notes and limitations: Extracted automatically from a single official source (section "Approval policies and sandbox modes"). Related: https://latestn.com/tool/codex/permissions-and-security/approve-requests-mobile How do I approve Codex action requests from my phone? https://latestn.com/tool/codex/permissions-and-security/auto-review-blocked-actions What does Codex Auto-review block and what does the reviewer see? https://latestn.com/tool/codex/permissions-and-security/bedrock-authentication How do I authenticate Codex Security scans with Amazon Bedrock? https://latestn.com/tool/codex/permissions-and-security/allow-only-approved-mcp-servers How do I allow only approved MCP servers in Codex? https://latestn.com/tool/codex/permissions-and-security/allowed-http-methods Which HTTP methods can I allow for Codex Cloud (Legacy) agent internet access? https://latestn.com/tool/codex/permissions-and-security/api-key-auth-automation How do I authenticate Codex in automation using an API key? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-17T13:59:32Z | changed | auto-review setting: — → auto_review | [1] 2026-11-17T13:59:32Z | changed | granular approval policy: — → approval_policy = { granular = { ... } } | [1] 2026-11-14T13:33:15Z | changed | auto-review setting: auto_review → — | [1] 2026-11-14T13:33:15Z | changed | granular approval policy: approval_policy = { granular = { ... } } → — | [1] 2026-11-13T03:48:19Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/codex/permissions-and-security/approval-policies-sandbox-modes/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/codex/permissions-and-security/approval-policies-sandbox-modes POST: https://latestn.com/dispute/tool/codex/permissions-and-security/approval-policies-sandbox-modes with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/codex/permissions-and-security/approval-policies-sandbox-modes?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). How do I configure approval policies and sandbox modes in Codex? Revision 3, last verified 2026-11-20T13:35:58Z. https://latestn.com/tool/codex/permissions-and-security/approval-policies-sandbox-modes (accessed 2026-11-21T10:08:28Z). @misc{LatestN_tool_codex_permissions_and_security_approval_policies_sandbox_modes_r3, title = {How do I configure approval policies and sandbox modes in Codex?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/codex/permissions-and-security/approval-policies-sandbox-modes}}, note = {Revision 3; digest sha256:c5c4102d3ba233c9; last verified 2026-11-20T13:35:58Z; accessed 2026-11-21T10:08:28Z} } 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/codex/permissions-and-security/approval-policies-sandbox-modes