config-reference.md Codex Docs, 2026-02-24 00:33 UTC → 2026-05-14 07:00 UTC

6 6 

7User-level configuration lives in `~/.codex/config.toml`. You can also add project-scoped overrides in `.codex/config.toml` files. Codex loads project-scoped config files only when you trust the project.7User-level configuration lives in `~/.codex/config.toml`. You can also add project-scoped overrides in `.codex/config.toml` files. Codex loads project-scoped config files only when you trust the project.

8 8 

9For sandbox and approval keys (`approval_policy`, `sandbox_mode`, and `sandbox_workspace_write.*`), pair this reference with [Sandbox and approvals](https://developers.openai.com/codex/security#sandbox-and-approvals), [Protected paths in writable roots](https://developers.openai.com/codex/security#protected-paths-in-writable-roots), and [Network access](https://developers.openai.com/codex/security#network-access).9For sandbox and approval keys (`approval_policy`, `sandbox_mode`, and `sandbox_workspace_write.*`), pair this reference with [Sandbox and approvals](https://developers.openai.com/codex/agent-approvals-security#sandbox-and-approvals), [Protected paths in writable roots](https://developers.openai.com/codex/agent-approvals-security#protected-paths-in-writable-roots), and [Network access](https://developers.openai.com/codex/agent-approvals-security#network-access).

10 10 

11| Key | Type / Values | Details |11<ConfigTable

12| --- | --- | --- |12 options={[

13| `agents.<name>.config_file` | `string (path)` | Path to a TOML config layer for that role; relative paths resolve from the config file that declares the role. |13 {

14| `agents.<name>.description` | `string` | Role guidance shown to Codex when choosing and spawning that agent type. |14 key: "model",

15| `agents.max_depth` | `number` | Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1). |15 type: "string",

16| `agents.max_threads` | `number` | Maximum number of agent threads that can be open concurrently. |16 description: "Model to use (e.g., `gpt-5.5`).",

17| `allow_login_shell` | `boolean` | Allow shell-based tools to use login-shell semantics. Defaults to `true`; when `false`, `login = true` requests are rejected and omitted `login` defaults to non-login shells. |17 },

18| `approval_policy` | `untrusted | on-request | never | { reject = { sandbox_approval = bool, rules = bool, mcp_elicitations = bool } }` | Controls when Codex pauses for approval before executing commands. You can also use `approval_policy = { reject = { ... } }` to auto-reject specific prompt categories while keeping other prompts interactive. `on-failure` is deprecated; use `on-request` for interactive runs or `never` for non-interactive runs. |18 {

19| `approval_policy.reject.mcp_elicitations` | `boolean` | When `true`, MCP elicitation prompts are auto-rejected instead of shown to the user. |19 key: "review_model",

20| `approval_policy.reject.rules` | `boolean` | When `true`, approvals triggered by execpolicy `prompt` rules are auto-rejected. |20 type: "string",

21| `approval_policy.reject.sandbox_approval` | `boolean` | When `true`, sandbox escalation approval prompts are auto-rejected. |21 description:

22| `apps._default.destructive_enabled` | `boolean` | Default allow/deny for app tools with `destructive_hint = true`. |22 "Optional model override used by `/review` (defaults to the current session model).",

23| `apps._default.enabled` | `boolean` | Default app enabled state for all apps unless overridden per app. |23 },

24| `apps._default.open_world_enabled` | `boolean` | Default allow/deny for app tools with `open_world_hint = true`. |24 {

25| `apps.<id>.default_tools_approval_mode` | `auto | prompt | approve` | Default approval behavior for tools in this app unless a per-tool override exists. |25 key: "model_provider",

26| `apps.<id>.default_tools_enabled` | `boolean` | Default enabled state for tools in this app unless a per-tool override exists. |26 type: "string",

27| `apps.<id>.destructive_enabled` | `boolean` | Allow or block tools in this app that advertise `destructive_hint = true`. |27 description: "Provider id from `model_providers` (default: `openai`).",

28| `apps.<id>.enabled` | `boolean` | Enable or disable a specific app/connector by id (default: true). |28 },

29| `apps.<id>.open_world_enabled` | `boolean` | Allow or block tools in this app that advertise `open_world_hint = true`. |29 {

30| `apps.<id>.tools.<tool>.approval_mode` | `auto | prompt | approve` | Per-tool approval behavior override for a single app tool. |30 key: "openai_base_url",

31| `apps.<id>.tools.<tool>.enabled` | `boolean` | Per-tool enabled override for an app tool (for example `repos/list`). |31 type: "string",

32| `background_terminal_max_timeout` | `number` | Maximum poll window in milliseconds for empty `write_stdin` polls (background terminal polling). Default: `300000` (5 minutes). Replaces the older `background_terminal_timeout` key. |32 description:

33| `chatgpt_base_url` | `string` | Override the base URL used during the ChatGPT login flow. |33 "Base URL override for the built-in `openai` model provider.",

34| `check_for_update_on_startup` | `boolean` | Check for Codex updates on startup (set to false only when updates are centrally managed). |34 },

35| `cli_auth_credentials_store` | `file | keyring | auto` | Control where the CLI stores cached credentials (file-based auth.json vs OS keychain). |35 {

36| `compact_prompt` | `string` | Inline override for the history compaction prompt. |36 key: "model_context_window",

37| `developer_instructions` | `string` | Additional developer instructions injected into the session (optional). |37 type: "number",

38| `disable_paste_burst` | `boolean` | Disable burst-paste detection in the TUI. |38 description: "Context window tokens available to the active model.",

39| `experimental_compact_prompt_file` | `string (path)` | Load the compaction prompt override from a file (experimental). |39 },

40| `experimental_use_freeform_apply_patch` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform` or `codex --enable apply_patch_freeform`. |40 {

41| `experimental_use_unified_exec_tool` | `boolean` | Legacy name for enabling unified exec; prefer `[features].unified_exec` or `codex --enable unified_exec`. |41 key: "model_auto_compact_token_limit",

42| `features.apply_patch_freeform` | `boolean` | Expose the freeform `apply_patch` tool (experimental). |42 type: "number",

43| `features.apps` | `boolean` | Enable ChatGPT Apps/connectors support (experimental). |43 description:

44| `features.apps_mcp_gateway` | `boolean` | Route Apps MCP calls through the OpenAI connectors MCP gateway (`https://api.openai.com/v1/connectors/mcp/`) instead of legacy routing (experimental). |44 "Token threshold that triggers automatic history compaction (unset uses model defaults).",

45| `features.child_agents_md` | `boolean` | Append AGENTS.md scope/precedence guidance even when no AGENTS.md is present (experimental). |45 },

46| `features.collaboration_modes` | `boolean` | Enable collaboration modes such as plan mode (stable; on by default). |46 {

47| `features.elevated_windows_sandbox` | `boolean` | Enable the elevated Windows sandbox pipeline (experimental). |47 key: "model_catalog_json",

48| `features.experimental_windows_sandbox` | `boolean` | Run the Windows restricted-token sandbox (experimental). |48 type: "string (path)",

49| `features.multi_agent` | `boolean` | Enable multi-agent collaboration tools (`spawn\_agent`, `send\_input`, `resume\_agent`, `wait`, and `close\_agent`) (experimental; off by default). |49 description:

50| `features.personality` | `boolean` | Enable personality selection controls (stable; on by default). |50 "Optional path to a JSON model catalog loaded on startup. Profile-level `profiles.<name>.model_catalog_json` can override this per profile.",

51| `features.powershell_utf8` | `boolean` | Force PowerShell UTF-8 output (defaults to true). |51 },

52| `features.remote_models` | `boolean` | Refresh remote model list before showing readiness (experimental). |52 {

53| `features.request_rule` | `boolean` | Enable Smart approvals (`prefix_rule` suggestions on escalation requests; stable; on by default). |53 key: "oss_provider",

54| `features.runtime_metrics` | `boolean` | Show runtime metrics summary in TUI turn separators (experimental). |54 type: "lmstudio | ollama",

55| `features.search_tool` | `boolean` | Enable `search_tool_bm25` for Apps tool discovery before invoking app MCP tools (experimental). |55 description:

56| `features.shell_snapshot` | `boolean` | Snapshot shell environment to speed up repeated commands (beta). |56 "Default local provider used when running with `--oss` (defaults to prompting if unset).",

57| `features.shell_tool` | `boolean` | Enable the default `shell` tool for running commands (stable; on by default). |57 },

58| `features.unified_exec` | `boolean` | Use the unified PTY-backed exec tool (beta). |58 {

59| `features.use_linux_sandbox_bwrap` | `boolean` | Use the bubblewrap-based Linux sandbox pipeline (experimental; off by default). |59 key: "approval_policy",

60| `features.web_search` | `boolean` | Deprecated legacy toggle; prefer the top-level `web_search` setting. |60 type: "untrusted | on-request | never | { granular = { sandbox_approval = bool, rules = bool, mcp_elicitations = bool, request_permissions = bool, skill_approval = bool } }",

61| `features.web_search_cached` | `boolean` | Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`. |61 description:

62| `features.web_search_request` | `boolean` | Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`. |62 "Controls when Codex pauses for approval before executing commands. You can also use `approval_policy = { granular = { ... } }` to allow or auto-reject specific prompt categories while keeping other prompts interactive. `on-failure` is deprecated; use `on-request` for interactive runs or `never` for non-interactive runs.",

63| `feedback.enabled` | `boolean` | Enable feedback submission via `/feedback` across Codex surfaces (default: true). |63 },

64| `file_opener` | `vscode | vscode-insiders | windsurf | cursor | none` | URI scheme used to open citations from Codex output (default: `vscode`). |64 {

65| `forced_chatgpt_workspace_id` | `string (uuid)` | Limit ChatGPT logins to a specific workspace identifier. |65 key: "approval_policy.granular.sandbox_approval",

66| `forced_login_method` | `chatgpt | api` | Restrict Codex to a specific authentication method. |66 type: "boolean",

67| `hide_agent_reasoning` | `boolean` | Suppress reasoning events in both the TUI and `codex exec` output. |67 description:

68| `history.max_bytes` | `number` | If set, caps the history file size in bytes by dropping oldest entries. |68 "When `true`, sandbox escalation approval prompts are allowed to surface.",

69| `history.persistence` | `save-all | none` | Control whether Codex saves session transcripts to history.jsonl. |69 },

70| `include_apply_patch_tool` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |70 {

71| `instructions` | `string` | Reserved for future use; prefer `model_instructions_file` or `AGENTS.md`. |71 key: "approval_policy.granular.rules",

72| `log_dir` | `string (path)` | Directory where Codex writes log files (for example `codex-tui.log`); defaults to `$CODEX_HOME/log`. |72 type: "boolean",

73| `mcp_oauth_callback_port` | `integer` | Optional fixed port for the local HTTP callback server used during MCP OAuth login. When unset, Codex binds to an ephemeral port chosen by the OS. |73 description:

74| `mcp_oauth_callback_url` | `string` | Optional redirect URI override for MCP OAuth login (for example, a devbox ingress URL). `mcp_oauth_callback_port` still controls the callback listener port. |74 "When `true`, approvals triggered by execpolicy `prompt` rules are allowed to surface.",

75| `mcp_oauth_credentials_store` | `auto | file | keyring` | Preferred store for MCP OAuth credentials. |75 },

76| `mcp_servers.<id>.args` | `array<string>` | Arguments passed to the MCP stdio server command. |76 {

77| `mcp_servers.<id>.bearer_token_env_var` | `string` | Environment variable sourcing the bearer token for an MCP HTTP server. |77 key: "approval_policy.granular.mcp_elicitations",

78| `mcp_servers.<id>.command` | `string` | Launcher command for an MCP stdio server. |78 type: "boolean",

79| `mcp_servers.<id>.cwd` | `string` | Working directory for the MCP stdio server process. |79 description:

80| `mcp_servers.<id>.disabled_tools` | `array<string>` | Deny list applied after `enabled_tools` for the MCP server. |80 "When `true`, MCP elicitation prompts are allowed to surface instead of being auto-rejected.",

81| `mcp_servers.<id>.enabled` | `boolean` | Disable an MCP server without removing its configuration. |81 },

82| `mcp_servers.<id>.enabled_tools` | `array<string>` | Allow list of tool names exposed by the MCP server. |82 {

83| `mcp_servers.<id>.env` | `map<string,string>` | Environment variables forwarded to the MCP stdio server. |83 key: "approval_policy.granular.request_permissions",

84| `mcp_servers.<id>.env_http_headers` | `map<string,string>` | HTTP headers populated from environment variables for an MCP HTTP server. |84 type: "boolean",

85| `mcp_servers.<id>.env_vars` | `array<string>` | Additional environment variables to whitelist for an MCP stdio server. |85 description:

86| `mcp_servers.<id>.http_headers` | `map<string,string>` | Static HTTP headers included with each MCP HTTP request. |86 "When `true`, prompts from the `request_permissions` tool are allowed to surface.",

87| `mcp_servers.<id>.required` | `boolean` | When true, fail startup/resume if this enabled MCP server cannot initialize. |87 },

88| `mcp_servers.<id>.startup_timeout_ms` | `number` | Alias for `startup_timeout_sec` in milliseconds. |88 {

89| `mcp_servers.<id>.startup_timeout_sec` | `number` | Override the default 10s startup timeout for an MCP server. |89 key: "approval_policy.granular.skill_approval",

90| `mcp_servers.<id>.tool_timeout_sec` | `number` | Override the default 60s per-tool timeout for an MCP server. |90 type: "boolean",

91| `mcp_servers.<id>.url` | `string` | Endpoint for an MCP streamable HTTP server. |91 description:

92| `model` | `string` | Model to use (e.g., `gpt-5-codex`). |92 "When `true`, skill-script approval prompts are allowed to surface.",

93| `model_auto_compact_token_limit` | `number` | Token threshold that triggers automatic history compaction (unset uses model defaults). |93 },

94| `model_catalog_json` | `string (path)` | Optional path to a JSON model catalog loaded on startup. Profile-level `profiles.<name>.model_catalog_json` can override this per profile. |94 {

95| `model_context_window` | `number` | Context window tokens available to the active model. |95 key: "approvals_reviewer",

96| `model_instructions_file` | `string (path)` | Replacement for built-in instructions instead of `AGENTS.md`. |96 type: "user | auto_review",

97| `model_provider` | `string` | Provider id from `model_providers` (default: `openai`). |97 description:

98| `model_providers.<id>.base_url` | `string` | API base URL for the model provider. |98 "Who reviews eligible approval prompts under `on-request` or granular approval policies. Defaults to `user`; `auto_review` uses the reviewer subagent. This setting doesn't change sandboxing or review actions already allowed inside the sandbox.",

99| `model_providers.<id>.env_http_headers` | `map<string,string>` | HTTP headers populated from environment variables when present. |99 },

100| `model_providers.<id>.env_key` | `string` | Environment variable supplying the provider API key. |100 {

101| `model_providers.<id>.env_key_instructions` | `string` | Optional setup guidance for the provider API key. |101 key: "auto_review.policy",

102| `model_providers.<id>.experimental_bearer_token` | `string` | Direct bearer token for the provider (discouraged; use `env_key`). |102 type: "string",

103| `model_providers.<id>.http_headers` | `map<string,string>` | Static HTTP headers added to provider requests. |103 description:

104| `model_providers.<id>.name` | `string` | Display name for a custom model provider. |104 "Local Markdown policy instructions for automatic review. Managed `guardian_policy_config` takes precedence. Blank values are ignored.",

105| `model_providers.<id>.query_params` | `map<string,string>` | Extra query parameters appended to provider requests. |105 },

106| `model_providers.<id>.request_max_retries` | `number` | Retry count for HTTP requests to the provider (default: 4). |106 {

107| `model_providers.<id>.requires_openai_auth` | `boolean` | The provider uses OpenAI authentication (defaults to false). |107 key: "allow_login_shell",

108| `model_providers.<id>.stream_idle_timeout_ms` | `number` | Idle timeout for SSE streams in milliseconds (default: 300000). |108 type: "boolean",

109| `model_providers.<id>.stream_max_retries` | `number` | Retry count for SSE streaming interruptions (default: 5). |109 description:

110| `model_providers.<id>.wire_api` | `chat | responses` | Protocol used by the provider (defaults to `chat` if omitted). |110 "Allow shell-based tools to use login-shell semantics. Defaults to `true`; when `false`, `login = true` requests are rejected and omitted `login` defaults to non-login shells.",

111| `model_reasoning_effort` | `minimal | low | medium | high | xhigh` | Adjust reasoning effort for supported models (Responses API only; `xhigh` is model-dependent). |111 },

112| `model_reasoning_summary` | `auto | concise | detailed | none` | Select reasoning summary detail or disable summaries entirely. |112 {

113| `model_supports_reasoning_summaries` | `boolean` | Force Codex to send or not send reasoning metadata. |113 key: "sandbox_mode",

114| `model_verbosity` | `low | medium | high` | Control GPT-5 Responses API verbosity (defaults to `medium`). |114 type: "read-only | workspace-write | danger-full-access",

115| `notice.hide_full_access_warning` | `boolean` | Track acknowledgement of the full access warning prompt. |115 description:

116| `notice.hide_gpt-5.1-codex-max_migration_prompt` | `boolean` | Track acknowledgement of the gpt-5.1-codex-max migration prompt. |116 "Sandbox policy for filesystem and network access during command execution.",

117| `notice.hide_gpt5_1_migration_prompt` | `boolean` | Track acknowledgement of the GPT-5.1 migration prompt. |117 },

118| `notice.hide_rate_limit_model_nudge` | `boolean` | Track opt-out of the rate limit model switch reminder. |118 {

119| `notice.hide_world_writable_warning` | `boolean` | Track acknowledgement of the Windows world-writable directories warning. |119 key: "sandbox_workspace_write.writable_roots",

120| `notice.model_migrations` | `map<string,string>` | Track acknowledged model migrations as old->new mappings. |120 type: "array<string>",

121| `notify` | `array<string>` | Command invoked for notifications; receives a JSON payload from Codex. |121 description:

122| `oss_provider` | `lmstudio | ollama` | Default local provider used when running with `--oss` (defaults to prompting if unset). |122 'Additional writable roots when `sandbox_mode = "workspace-write"`.',

123| `otel.environment` | `string` | Environment tag applied to emitted OpenTelemetry events (default: `dev`). |123 },

124| `otel.exporter` | `none | otlp-http | otlp-grpc` | Select the OpenTelemetry exporter and provide any endpoint metadata. |124 {

125| `otel.exporter.<id>.endpoint` | `string` | Exporter endpoint for OTEL logs. |125 key: "sandbox_workspace_write.network_access",

126| `otel.exporter.<id>.headers` | `map<string,string>` | Static headers included with OTEL exporter requests. |126 type: "boolean",

127| `otel.exporter.<id>.protocol` | `binary | json` | Protocol used by the OTLP/HTTP exporter. |127 description:

128| `otel.exporter.<id>.tls.ca-certificate` | `string` | CA certificate path for OTEL exporter TLS. |128 "Allow outbound network access inside the workspace-write sandbox.",

129| `otel.exporter.<id>.tls.client-certificate` | `string` | Client certificate path for OTEL exporter TLS. |129 },

130| `otel.exporter.<id>.tls.client-private-key` | `string` | Client private key path for OTEL exporter TLS. |130 {

131| `otel.log_user_prompt` | `boolean` | Opt in to exporting raw user prompts with OpenTelemetry logs. |131 key: "sandbox_workspace_write.exclude_tmpdir_env_var",

132| `otel.trace_exporter` | `none | otlp-http | otlp-grpc` | Select the OpenTelemetry trace exporter and provide any endpoint metadata. |132 type: "boolean",

133| `otel.trace_exporter.<id>.endpoint` | `string` | Trace exporter endpoint for OTEL logs. |133 description:

134| `otel.trace_exporter.<id>.headers` | `map<string,string>` | Static headers included with OTEL trace exporter requests. |134 "Exclude `$TMPDIR` from writable roots in workspace-write mode.",

135| `otel.trace_exporter.<id>.protocol` | `binary | json` | Protocol used by the OTLP/HTTP trace exporter. |135 },

136| `otel.trace_exporter.<id>.tls.ca-certificate` | `string` | CA certificate path for OTEL trace exporter TLS. |136 {

137| `otel.trace_exporter.<id>.tls.client-certificate` | `string` | Client certificate path for OTEL trace exporter TLS. |137 key: "sandbox_workspace_write.exclude_slash_tmp",

138| `otel.trace_exporter.<id>.tls.client-private-key` | `string` | Client private key path for OTEL trace exporter TLS. |138 type: "boolean",

139| `personality` | `none | friendly | pragmatic` | Default communication style for models that advertise `supportsPersonality`; can be overridden per thread/turn or via `/personality`. |139 description:

140| `profile` | `string` | Default profile applied at startup (equivalent to `--profile`). |140 "Exclude `/tmp` from writable roots in workspace-write mode.",

141| `profiles.<name>.*` | `various` | Profile-scoped overrides for any of the supported configuration keys. |141 },

142| `profiles.<name>.experimental_use_freeform_apply_patch` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |142 {

143| `profiles.<name>.experimental_use_unified_exec_tool` | `boolean` | Legacy name for enabling unified exec; prefer `[features].unified_exec`. |143 key: "windows.sandbox",

144| `profiles.<name>.include_apply_patch_tool` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |144 type: "unelevated | elevated",

145| `profiles.<name>.model_catalog_json` | `string (path)` | Profile-scoped model catalog JSON path override (applied on startup only; overrides the top-level `model_catalog_json` for that profile). |145 description:

146| `profiles.<name>.oss_provider` | `lmstudio | ollama` | Profile-scoped OSS provider for `--oss` sessions. |146 "Windows-only native sandbox mode when running Codex natively on Windows.",

147| `profiles.<name>.personality` | `none | friendly | pragmatic` | Profile-scoped communication style override for supported models. |147 },

148| `profiles.<name>.web_search` | `disabled | cached | live` | Profile-scoped web search mode override (default: `"cached"`). |148 {

149| `project_doc_fallback_filenames` | `array<string>` | Additional filenames to try when `AGENTS.md` is missing. |149 key: "windows.sandbox_private_desktop",

150| `project_doc_max_bytes` | `number` | Maximum bytes read from `AGENTS.md` when building project instructions. |150 type: "boolean",

151| `project_root_markers` | `array<string>` | List of project root marker filenames; used when searching parent directories for the project root. |151 description:

152| `projects.<path>.trust_level` | `string` | Mark a project or worktree as trusted or untrusted (`"trusted"` | `"untrusted"`). Untrusted projects skip project-scoped `.codex/` layers. |152 "Run the final sandboxed child process on a private desktop by default on native Windows. Set `false` only for compatibility with the older `Winsta0\\\\Default` behavior.",

153| `review_model` | `string` | Optional model override used by `/review` (defaults to the current session model). |153 },

154| `sandbox_mode` | `read-only | workspace-write | danger-full-access` | Sandbox policy for filesystem and network access during command execution. |154 {

155| `sandbox_workspace_write.exclude_slash_tmp` | `boolean` | Exclude `/tmp` from writable roots in workspace-write mode. |155 key: "notify",

156| `sandbox_workspace_write.exclude_tmpdir_env_var` | `boolean` | Exclude `$TMPDIR` from writable roots in workspace-write mode. |156 type: "array<string>",

157| `sandbox_workspace_write.network_access` | `boolean` | Allow outbound network access inside the workspace-write sandbox. |157 description:

158| `sandbox_workspace_write.writable_roots` | `array<string>` | Additional writable roots when `sandbox_mode = "workspace-write"`. |158 "Command invoked for notifications; receives a JSON payload from Codex.",

159| `shell_environment_policy.exclude` | `array<string>` | Glob patterns for removing environment variables after the defaults. |159 },

160| `shell_environment_policy.experimental_use_profile` | `boolean` | Use the user shell profile when spawning subprocesses. |160 {

161| `shell_environment_policy.ignore_default_excludes` | `boolean` | Keep variables containing KEY/SECRET/TOKEN before other filters run. |161 key: "check_for_update_on_startup",

162| `shell_environment_policy.include_only` | `array<string>` | Whitelist of patterns; when set only matching variables are kept. |162 type: "boolean",

163| `shell_environment_policy.inherit` | `all | core | none` | Baseline environment inheritance when spawning subprocesses. |163 description:

164| `shell_environment_policy.set` | `map<string,string>` | Explicit environment overrides injected into every subprocess. |164 "Check for Codex updates on startup (set to false only when updates are centrally managed).",

165| `show_raw_agent_reasoning` | `boolean` | Surface raw reasoning content when the active model emits it. |165 },

166| `skills.config` | `array<object>` | Per-skill enablement overrides stored in config.toml. |166 {

167| `skills.config.<index>.enabled` | `boolean` | Enable or disable the referenced skill. |167 key: "feedback.enabled",

168| `skills.config.<index>.path` | `string (path)` | Path to a skill folder containing `SKILL.md`. |168 type: "boolean",

169| `suppress_unstable_features_warning` | `boolean` | Suppress the warning that appears when under-development feature flags are enabled. |169 description:

170| `tool_output_token_limit` | `number` | Token budget for storing individual tool/function outputs in history. |170 "Enable feedback submission via `/feedback` across Codex surfaces (default: true).",

171| `tools.web_search` | `boolean` | Deprecated legacy toggle for web search; prefer the top-level `web_search` setting. |171 },

172| `tui` | `table` | TUI-specific options such as enabling inline desktop notifications. |172 {

173| `tui.alternate_screen` | `auto | always | never` | Control alternate screen usage for the TUI (default: auto; auto skips it in Zellij to preserve scrollback). |173 key: "analytics.enabled",

174| `tui.animations` | `boolean` | Enable terminal animations (welcome screen, shimmer, spinner) (default: true). |174 type: "boolean",

175| `tui.notification_method` | `auto | osc9 | bel` | Notification method for unfocused terminal notifications (default: auto). |175 description:

176| `tui.notifications` | `boolean | array<string>` | Enable TUI notifications; optionally restrict to specific event types. |176 "Enable or disable analytics for this machine/profile. When unset, the client default applies.",

177| `tui.show_tooltips` | `boolean` | Show onboarding tooltips in the TUI welcome screen (default: true). |177 },

178| `tui.status_line` | `array<string> | null` | Ordered list of TUI footer status-line item identifiers. `null` disables the status line. |178 {

179| `web_search` | `disabled | cached | live` | Web search mode (default: `"cached"`; cached uses an OpenAI-maintained index and does not fetch live pages; if you use `--yolo` or another full access sandbox setting, it defaults to `"live"`). Use `"live"` to fetch the most recent data from the web, or `"disabled"` to remove the tool. |179 key: "instructions",

180| `windows_wsl_setup_acknowledged` | `boolean` | Track Windows onboarding acknowledgement (Windows only). |180 type: "string",

181 181 description:

182Key182 "Reserved for future use; prefer `model_instructions_file` or `AGENTS.md`.",

183 183 },

184`agents.<name>.config_file`184 {

185 185 key: "developer_instructions",

186Type / Values186 type: "string",

187 187 description:

188`string (path)`188 "Additional developer instructions injected into the session (optional).",

189 189 },

190Details190 {

191 191 key: "log_dir",

192Path to a TOML config layer for that role; relative paths resolve from the config file that declares the role.192 type: "string (path)",

193 193 description:

194Key194 "Directory where Codex writes log files (for example `codex-tui.log`); defaults to `$CODEX_HOME/log`.",

195 195 },

196`agents.<name>.description`196 {

197 197 key: "sqlite_home",

198Type / Values198 type: "string (path)",

199 199 description:

200`string`200 "Directory where Codex stores the SQLite-backed state DB used by agent jobs and other resumable runtime state.",

201 201 },

202Details202 {

203 203 key: "compact_prompt",

204Role guidance shown to Codex when choosing and spawning that agent type.204 type: "string",

205 205 description: "Inline override for the history compaction prompt.",

206Key206 },

207 207 {

208`agents.max_depth`208 key: "commit_attribution",

209 209 type: "string",

210Type / Values210 description:

211 211 'Commit co-author trailer used when `[features].codex_git_commit` is enabled. Defaults to `Codex <noreply@openai.com>`; set `""` to disable.',

212`number`212 },

213 213 {

214Details214 key: "model_instructions_file",

215 215 type: "string (path)",

216Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1).216 description:

217 217 "Replacement for built-in instructions instead of `AGENTS.md`.",

218Key218 },

219 219 {

220`agents.max_threads`220 key: "personality",

221 221 type: "none | friendly | pragmatic",

222Type / Values222 description:

223 223 "Default communication style for models that advertise `supportsPersonality`; can be overridden per thread/turn or via `/personality`.",

224`number`224 },

225 225 {

226Details226 key: "service_tier",

227 227 type: "flex | fast",

228Maximum number of agent threads that can be open concurrently.228 description: "Preferred service tier for new turns.",

229 229 },

230Key230 {

231 231 key: "experimental_compact_prompt_file",

232`allow_login_shell`232 type: "string (path)",

233 233 description:

234Type / Values234 "Load the compaction prompt override from a file (experimental).",

235 235 },

236`boolean`236 {

237 237 key: "skills.config",

238Details238 type: "array<object>",

239 239 description: "Per-skill enablement overrides stored in config.toml.",

240Allow shell-based tools to use login-shell semantics. Defaults to `true`; when `false`, `login = true` requests are rejected and omitted `login` defaults to non-login shells.240 },

241 241 {

242Key242 key: "skills.config.<index>.path",

243 243 type: "string (path)",

244`approval_policy`244 description: "Path to a skill folder containing `SKILL.md`.",

245 245 },

246Type / Values246 {

247 247 key: "skills.config.<index>.enabled",

248`untrusted | on-request | never | { reject = { sandbox_approval = bool, rules = bool, mcp_elicitations = bool } }`248 type: "boolean",

249 249 description: "Enable or disable the referenced skill.",

250Details250 },

251 251 {

252Controls when Codex pauses for approval before executing commands. You can also use `approval_policy = { reject = { ... } }` to auto-reject specific prompt categories while keeping other prompts interactive. `on-failure` is deprecated; use `on-request` for interactive runs or `never` for non-interactive runs.252 key: "apps.<id>.enabled",

253 253 type: "boolean",

254Key254 description:

255 255 "Enable or disable a specific app/connector by id (default: true).",

256`approval_policy.reject.mcp_elicitations`256 },

257 257 {

258Type / Values258 key: "apps._default.enabled",

259 259 type: "boolean",

260`boolean`260 description:

261 261 "Default app enabled state for all apps unless overridden per app.",

262Details262 },

263 263 {

264When `true`, MCP elicitation prompts are auto-rejected instead of shown to the user.264 key: "apps._default.destructive_enabled",

265 265 type: "boolean",

266Key266 description:

267 267 "Default allow/deny for app tools with `destructive_hint = true`.",

268`approval_policy.reject.rules`268 },

269 269 {

270Type / Values270 key: "apps._default.open_world_enabled",

271 271 type: "boolean",

272`boolean`272 description:

273 273 "Default allow/deny for app tools with `open_world_hint = true`.",

274Details274 },

275 275 {

276When `true`, approvals triggered by execpolicy `prompt` rules are auto-rejected.276 key: "apps.<id>.destructive_enabled",

277 277 type: "boolean",

278Key278 description:

279 279 "Allow or block tools in this app that advertise `destructive_hint = true`.",

280`approval_policy.reject.sandbox_approval`280 },

281 281 {

282Type / Values282 key: "apps.<id>.open_world_enabled",

283 283 type: "boolean",

284`boolean`284 description:

285 285 "Allow or block tools in this app that advertise `open_world_hint = true`.",

286Details286 },

287 287 {

288When `true`, sandbox escalation approval prompts are auto-rejected.288 key: "apps.<id>.default_tools_enabled",

289 289 type: "boolean",

290Key290 description:

291 291 "Default enabled state for tools in this app unless a per-tool override exists.",

292`apps._default.destructive_enabled`292 },

293 293 {

294Type / Values294 key: "apps.<id>.default_tools_approval_mode",

295 295 type: "auto | prompt | approve",

296`boolean`296 description:

297 297 "Default approval behavior for tools in this app unless a per-tool override exists.",

298Details298 },

299 299 {

300Default allow/deny for app tools with `destructive_hint = true`.300 key: "apps.<id>.tools.<tool>.enabled",

301 301 type: "boolean",

302Key302 description:

303 303 "Per-tool enabled override for an app tool (for example `repos/list`).",

304`apps._default.enabled`304 },

305 305 {

306Type / Values306 key: "apps.<id>.tools.<tool>.approval_mode",

307 307 type: "auto | prompt | approve",

308`boolean`308 description: "Per-tool approval behavior override for a single app tool.",

309 309 },

310Details310 {

311 311 key: "tool_suggest.discoverables",

312Default app enabled state for all apps unless overridden per app.312 type: "array<table>",

313 313 description:

314Key314 'Allow tool suggestions for additional discoverable connectors or plugins. Each entry uses `type = "connector"` or `"plugin"` and an `id`.',

315 315 },

316`apps._default.open_world_enabled`316 {

317 317 key: "tool_suggest.disabled_tools",

318Type / Values318 type: "array<table>",

319 319 description:

320`boolean`320 'Disable suggestions for specific discoverable connectors or plugins. Each entry uses `type = "connector"` or `"plugin"` and an `id`.',

321 321 },

322Details322 {

323 323 key: "features.apps",

324Default allow/deny for app tools with `open_world_hint = true`.324 type: "boolean",

325 325 description: "Enable ChatGPT Apps/connectors support (experimental).",

326Key326 },

327 327 {

328`apps.<id>.default_tools_approval_mode`328 key: "features.codex_hooks",

329 329 type: "boolean",

330Type / Values330 description:

331 331 "Enable lifecycle hooks loaded from `hooks.json` or inline `[hooks]` config.",

332`auto | prompt | approve`332 },

333 333 {

334Details334 key: "features.codex_git_commit",

335 335 type: "boolean",

336Default approval behavior for tools in this app unless a per-tool override exists.336 description:

337 337 "Enable Codex-generated git commits. When enabled, Codex uses `commit_attribution` to append a `Co-authored-by:` trailer to generated commit messages.",

338Key338 },

339 339 {

340`apps.<id>.default_tools_enabled`340 key: "hooks",

341 341 type: "table",

342Type / Values342 description:

343 343 "Lifecycle hooks configured inline in `config.toml`. Uses the same event schema as `hooks.json`; see the Hooks guide for examples and supported events.",

344`boolean`344 },

345 345 {

346Details346 key: "features.memories",

347 347 type: "boolean",

348Default enabled state for tools in this app unless a per-tool override exists.348 description: "Enable [Memories](https://developers.openai.com/codex/memories) (off by default).",

349 349 },

350Key350 {

351 351 key: "mcp_servers.<id>.command",

352`apps.<id>.destructive_enabled`352 type: "string",

353 353 description: "Launcher command for an MCP stdio server.",

354Type / Values354 },

355 355 {

356`boolean`356 key: "mcp_servers.<id>.args",

357 357 type: "array<string>",

358Details358 description: "Arguments passed to the MCP stdio server command.",

359 359 },

360Allow or block tools in this app that advertise `destructive_hint = true`.360 {

361 361 key: "mcp_servers.<id>.env",

362Key362 type: "map<string,string>",

363 363 description: "Environment variables forwarded to the MCP stdio server.",

364`apps.<id>.enabled`364 },

365 365 {

366Type / Values366 key: "mcp_servers.<id>.env_vars",

367 367 type: 'array<string | { name = string, source = "local" | "remote" }>',

368`boolean`368 description:

369 369 'Additional environment variables to whitelist for an MCP stdio server. String entries default to `source = "local"`; use `source = "remote"` only with executor-backed remote stdio.',

370Details370 },

371 371 {

372Enable or disable a specific app/connector by id (default: true).372 key: "mcp_servers.<id>.cwd",

373 373 type: "string",

374Key374 description: "Working directory for the MCP stdio server process.",

375 375 },

376`apps.<id>.open_world_enabled`376 {

377 377 key: "mcp_servers.<id>.url",

378Type / Values378 type: "string",

379 379 description: "Endpoint for an MCP streamable HTTP server.",

380`boolean`380 },

381 381 {

382Details382 key: "mcp_servers.<id>.bearer_token_env_var",

383 383 type: "string",

384Allow or block tools in this app that advertise `open_world_hint = true`.384 description:

385 385 "Environment variable sourcing the bearer token for an MCP HTTP server.",

386Key386 },

387 387 {

388`apps.<id>.tools.<tool>.approval_mode`388 key: "mcp_servers.<id>.http_headers",

389 389 type: "map<string,string>",

390Type / Values390 description: "Static HTTP headers included with each MCP HTTP request.",

391 391 },

392`auto | prompt | approve`392 {

393 393 key: "mcp_servers.<id>.env_http_headers",

394Details394 type: "map<string,string>",

395 395 description:

396Per-tool approval behavior override for a single app tool.396 "HTTP headers populated from environment variables for an MCP HTTP server.",

397 397 },

398Key398 {

399 399 key: "mcp_servers.<id>.enabled",

400`apps.<id>.tools.<tool>.enabled`400 type: "boolean",

401 401 description: "Disable an MCP server without removing its configuration.",

402Type / Values402 },

403 403 {

404`boolean`404 key: "mcp_servers.<id>.required",

405 405 type: "boolean",

406Details406 description:

407 407 "When true, fail startup/resume if this enabled MCP server cannot initialize.",

408Per-tool enabled override for an app tool (for example `repos/list`).408 },

409 409 {

410Key410 key: "mcp_servers.<id>.startup_timeout_sec",

411 411 type: "number",

412`background_terminal_max_timeout`412 description:

413 413 "Override the default 10s startup timeout for an MCP server.",

414Type / Values414 },

415 415 {

416`number`416 key: "mcp_servers.<id>.startup_timeout_ms",

417 417 type: "number",

418Details418 description: "Alias for `startup_timeout_sec` in milliseconds.",

419 419 },

420Maximum poll window in milliseconds for empty `write_stdin` polls (background terminal polling). Default: `300000` (5 minutes). Replaces the older `background_terminal_timeout` key.420 {

421 421 key: "mcp_servers.<id>.tool_timeout_sec",

422Key422 type: "number",

423 423 description:

424`chatgpt_base_url`424 "Override the default 60s per-tool timeout for an MCP server.",

425 425 },

426Type / Values426 {

427 427 key: "mcp_servers.<id>.enabled_tools",

428`string`428 type: "array<string>",

429 429 description: "Allow list of tool names exposed by the MCP server.",

430Details430 },

431 431 {

432Override the base URL used during the ChatGPT login flow.432 key: "mcp_servers.<id>.disabled_tools",

433 433 type: "array<string>",

434Key434 description:

435 435 "Deny list applied after `enabled_tools` for the MCP server.",

436`check_for_update_on_startup`436 },

437 437 {

438Type / Values438 key: "mcp_servers.<id>.scopes",

439 439 type: "array<string>",

440`boolean`440 description:

441 441 "OAuth scopes to request when authenticating to that MCP server.",

442Details442 },

443 443 {

444Check for Codex updates on startup (set to false only when updates are centrally managed).444 key: "mcp_servers.<id>.oauth_resource",

445 445 type: "string",

446Key446 description:

447 447 "Optional RFC 8707 OAuth resource parameter to include during MCP login.",

448`cli_auth_credentials_store`448 },

449 449 {

450Type / Values450 key: "mcp_servers.<id>.experimental_environment",

451 451 type: "local | remote",

452`file | keyring | auto`452 description:

453 453 "Experimental placement for an MCP server. `remote` starts stdio servers through a remote executor environment; streamable HTTP remote placement is not implemented.",

454Details454 },

455 455 {

456Control where the CLI stores cached credentials (file-based auth.json vs OS keychain).456 key: "agents.max_threads",

457 457 type: "number",

458Key458 description:

459 459 "Maximum number of agent threads that can be open concurrently. Defaults to `6` when unset.",

460`compact_prompt`460 },

461 461 {

462Type / Values462 key: "agents.max_depth",

463 463 type: "number",

464`string`464 description:

465 465 "Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1).",

466Details466 },

467 467 {

468Inline override for the history compaction prompt.468 key: "agents.job_max_runtime_seconds",

469 469 type: "number",

470Key470 description:

471 471 "Default per-worker timeout for `spawn_agents_on_csv` jobs. When unset, the tool falls back to 1800 seconds per worker.",

472`developer_instructions`472 },

473 473 {

474Type / Values474 key: "agents.<name>.description",

475 475 type: "string",

476`string`476 description:

477 477 "Role guidance shown to Codex when choosing and spawning that agent type.",

478Details478 },

479 479 {

480Additional developer instructions injected into the session (optional).480 key: "agents.<name>.config_file",

481 481 type: "string (path)",

482Key482 description:

483 483 "Path to a TOML config layer for that role; relative paths resolve from the config file that declares the role.",

484`disable_paste_burst`484 },

485 485 {

486Type / Values486 key: "agents.<name>.nickname_candidates",

487 487 type: "array<string>",

488`boolean`488 description:

489 489 "Optional pool of display nicknames for spawned agents in that role.",

490Details490 },

491 491 {

492Disable burst-paste detection in the TUI.492 key: "memories.generate_memories",

493 493 type: "boolean",

494Key494 description:

495 495 "When `false`, newly created threads are not stored as memory-generation inputs. Defaults to `true`.",

496`experimental_compact_prompt_file`496 },

497 497 {

498Type / Values498 key: "memories.use_memories",

499 499 type: "boolean",

500`string (path)`500 description:

501 501 "When `false`, Codex skips injecting existing memories into future sessions. Defaults to `true`.",

502Details502 },

503 503 {

504Load the compaction prompt override from a file (experimental).504 key: "memories.disable_on_external_context",

505 505 type: "boolean",

506Key506 description:

507 507 "When `true`, threads that use external context such as MCP tool calls, web search, or tool search are kept out of memory generation. Defaults to `false`. Legacy alias: `memories.no_memories_if_mcp_or_web_search`.",

508`experimental_use_freeform_apply_patch`508 },

509 509 {

510Type / Values510 key: "memories.max_raw_memories_for_consolidation",

511 511 type: "number",

512`boolean`512 description:

513 513 "Maximum recent raw memories retained for global consolidation. Defaults to `256` and is capped at `4096`.",

514Details514 },

515 515 {

516Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform` or `codex --enable apply_patch_freeform`.516 key: "memories.max_unused_days",

517 517 type: "number",

518Key518 description:

519 519 "Maximum days since a memory was last used before it becomes ineligible for consolidation. Defaults to `30` and is clamped to `0`-`365`.",

520`experimental_use_unified_exec_tool`520 },

521 521 {

522Type / Values522 key: "memories.max_rollout_age_days",

523 523 type: "number",

524`boolean`524 description:

525 525 "Maximum age of threads considered for memory generation. Defaults to `30` and is clamped to `0`-`90`.",

526Details526 },

527 527 {

528Legacy name for enabling unified exec; prefer `[features].unified_exec` or `codex --enable unified_exec`.528 key: "memories.max_rollouts_per_startup",

529 529 type: "number",

530Key530 description:

531 531 "Maximum rollout candidates processed per startup pass. Defaults to `16` and is capped at `128`.",

532`features.apply_patch_freeform`532 },

533 533 {

534Type / Values534 key: "memories.min_rollout_idle_hours",

535 535 type: "number",

536`boolean`536 description:

537 537 "Minimum idle time before a thread is considered for memory generation. Defaults to `6` and is clamped to `1`-`48`.",

538Details538 },

539 539 {

540Expose the freeform `apply_patch` tool (experimental).540 key: "memories.min_rate_limit_remaining_percent",

541 541 type: "number",

542Key542 description:

543 543 "Minimum remaining percentage required in Codex rate-limit windows before memory generation starts. Defaults to `25` and is clamped to `0`-`100`.",

544`features.apps`544 },

545 545 {

546Type / Values546 key: "memories.extract_model",

547 547 type: "string",

548`boolean`548 description: "Optional model override for per-thread memory extraction.",

549 549 },

550Details550 {

551 551 key: "memories.consolidation_model",

552Enable ChatGPT Apps/connectors support (experimental).552 type: "string",

553 553 description: "Optional model override for global memory consolidation.",

554Key554 },

555 555 {

556`features.apps_mcp_gateway`556 key: "features.unified_exec",

557 557 type: "boolean",

558Type / Values558 description:

559 559 "Use the unified PTY-backed exec tool (stable; enabled by default except on Windows).",

560`boolean`560 },

561 561 {

562Details562 key: "features.shell_snapshot",

563 563 type: "boolean",

564Route Apps MCP calls through the OpenAI connectors MCP gateway (`https://api.openai.com/v1/connectors/mcp/`) instead of legacy routing (experimental).564 description:

565 565 "Snapshot shell environment to speed up repeated commands (stable; on by default).",

566Key566 },

567 567 {

568`features.child_agents_md`568 key: "features.undo",

569 569 type: "boolean",

570Type / Values570 description: "Enable undo support (stable; off by default).",

571 571 },

572`boolean`572 {

573 573 key: "features.multi_agent",

574Details574 type: "boolean",

575 575 description:

576Append AGENTS.md scope/precedence guidance even when no AGENTS.md is present (experimental).576 "Enable multi-agent collaboration tools (`spawn_agent`, `send_input`, `resume_agent`, `wait_agent`, and `close_agent`) (stable; on by default).",

577 577 },

578Key578 {

579 579 key: "features.personality",

580`features.collaboration_modes`580 type: "boolean",

581 581 description:

582Type / Values582 "Enable personality selection controls (stable; on by default).",

583 583 },

584`boolean`584 {

585 585 key: "features.web_search",

586Details586 type: "boolean",

587 587 description:

588Enable collaboration modes such as plan mode (stable; on by default).588 "Deprecated legacy toggle; prefer the top-level `web_search` setting.",

589 589 },

590Key590 {

591 591 key: "features.web_search_cached",

592`features.elevated_windows_sandbox`592 type: "boolean",

593 593 description:

594Type / Values594 'Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`.',

595 595 },

596`boolean`596 {

597 597 key: "features.web_search_request",

598Details598 type: "boolean",

599 599 description:

600Enable the elevated Windows sandbox pipeline (experimental).600 'Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`.',

601 601 },

602Key602 {

603 603 key: "features.shell_tool",

604`features.experimental_windows_sandbox`604 type: "boolean",

605 605 description:

606Type / Values606 "Enable the default `shell` tool for running commands (stable; on by default).",

607 607 },

608`boolean`608 {

609 609 key: "features.enable_request_compression",

610Details610 type: "boolean",

611 611 description:

612Run the Windows restricted-token sandbox (experimental).612 "Compress streaming request bodies with zstd when supported (stable; on by default).",

613 613 },

614Key614 {

615 615 key: "features.skill_mcp_dependency_install",

616`features.multi_agent`616 type: "boolean",

617 617 description:

618Type / Values618 "Allow prompting and installing missing MCP dependencies for skills (stable; on by default).",

619 619 },

620`boolean`620 {

621 621 key: "features.fast_mode",

622Details622 type: "boolean",

623 623 description:

624Enable multi-agent collaboration tools (`spawn\_agent`, `send\_input`, `resume\_agent`, `wait`, and `close\_agent`) (experimental; off by default).624 'Enable Fast mode selection and the `service_tier = "fast"` path (stable; on by default).',

625 625 },

626Key626 {

627 627 key: "features.prevent_idle_sleep",

628`features.personality`628 type: "boolean",

629 629 description:

630Type / Values630 "Prevent the machine from sleeping while a turn is actively running (experimental; off by default).",

631 631 },

632`boolean`632 {

633 633 key: "suppress_unstable_features_warning",

634Details634 type: "boolean",

635 635 description:

636Enable personality selection controls (stable; on by default).636 "Suppress the warning that appears when under-development feature flags are enabled.",

637 637 },

638Key638 {

639 639 key: "model_providers.<id>",

640`features.powershell_utf8`640 type: "table",

641 641 description:

642Type / Values642 "Custom provider definition. Built-in provider IDs (`openai`, `ollama`, and `lmstudio`) are reserved and cannot be overridden.",

643 643 },

644`boolean`644 {

645 645 key: "model_providers.<id>.name",

646Details646 type: "string",

647 647 description: "Display name for a custom model provider.",

648Force PowerShell UTF-8 output (defaults to true).648 },

649 649 {

650Key650 key: "model_providers.<id>.base_url",

651 651 type: "string",

652`features.remote_models`652 description: "API base URL for the model provider.",

653 653 },

654Type / Values654 {

655 655 key: "model_providers.<id>.env_key",

656`boolean`656 type: "string",

657 657 description: "Environment variable supplying the provider API key.",

658Details658 },

659 659 {

660Refresh remote model list before showing readiness (experimental).660 key: "model_providers.<id>.env_key_instructions",

661 661 type: "string",

662Key662 description: "Optional setup guidance for the provider API key.",

663 663 },

664`features.request_rule`664 {

665 665 key: "model_providers.<id>.experimental_bearer_token",

666Type / Values666 type: "string",

667 667 description:

668`boolean`668 "Direct bearer token for the provider (discouraged; use `env_key`).",

669 669 },

670Details670 {

671 671 key: "model_providers.<id>.requires_openai_auth",

672Enable Smart approvals (`prefix_rule` suggestions on escalation requests; stable; on by default).672 type: "boolean",

673 673 description:

674Key674 "The provider uses OpenAI authentication (defaults to false).",

675 675 },

676`features.runtime_metrics`676 {

677 677 key: "model_providers.<id>.wire_api",

678Type / Values678 type: "responses",

679 679 description:

680`boolean`680 "Protocol used by the provider. `responses` is the only supported value, and it is the default when omitted.",

681 681 },

682Details682 {

683 683 key: "model_providers.<id>.query_params",

684Show runtime metrics summary in TUI turn separators (experimental).684 type: "map<string,string>",

685 685 description: "Extra query parameters appended to provider requests.",

686Key686 },

687 687 {

688`features.search_tool`688 key: "model_providers.<id>.http_headers",

689 689 type: "map<string,string>",

690Type / Values690 description: "Static HTTP headers added to provider requests.",

691 691 },

692`boolean`692 {

693 693 key: "model_providers.<id>.env_http_headers",

694Details694 type: "map<string,string>",

695 695 description:

696Enable `search_tool_bm25` for Apps tool discovery before invoking app MCP tools (experimental).696 "HTTP headers populated from environment variables when present.",

697 697 },

698Key698 {

699 699 key: "model_providers.<id>.request_max_retries",

700`features.shell_snapshot`700 type: "number",

701 701 description:

702Type / Values702 "Retry count for HTTP requests to the provider (default: 4).",

703 703 },

704`boolean`704 {

705 705 key: "model_providers.<id>.stream_max_retries",

706Details706 type: "number",

707 707 description: "Retry count for SSE streaming interruptions (default: 5).",

708Snapshot shell environment to speed up repeated commands (beta).708 },

709 709 {

710Key710 key: "model_providers.<id>.stream_idle_timeout_ms",

711 711 type: "number",

712`features.shell_tool`712 description:

713 713 "Idle timeout for SSE streams in milliseconds (default: 300000).",

714Type / Values714 },

715 715 {

716`boolean`716 key: "model_providers.<id>.supports_websockets",

717 717 type: "boolean",

718Details718 description:

719 719 "Whether that provider supports the Responses API WebSocket transport.",

720Enable the default `shell` tool for running commands (stable; on by default).720 },

721 721 {

722Key722 key: "model_providers.<id>.auth",

723 723 type: "table",

724`features.unified_exec`724 description:

725 725 "Command-backed bearer token configuration for a custom provider. Do not combine with `env_key`, `experimental_bearer_token`, or `requires_openai_auth`.",

726Type / Values726 },

727 727 {

728`boolean`728 key: "model_providers.<id>.auth.command",

729 729 type: "string",

730Details730 description:

731 731 "Command to run when Codex needs a bearer token. The command must print the token to stdout.",

732Use the unified PTY-backed exec tool (beta).732 },

733 733 {

734Key734 key: "model_providers.<id>.auth.args",

735 735 type: "array<string>",

736`features.use_linux_sandbox_bwrap`736 description: "Arguments passed to the token command.",

737 737 },

738Type / Values738 {

739 739 key: "model_providers.<id>.auth.timeout_ms",

740`boolean`740 type: "number",

741 741 description:

742Details742 "Maximum token command runtime in milliseconds (default: 5000).",

743 743 },

744Use the bubblewrap-based Linux sandbox pipeline (experimental; off by default).744 {

745 745 key: "model_providers.<id>.auth.refresh_interval_ms",

746Key746 type: "number",

747 747 description:

748`features.web_search`748 "How often Codex proactively refreshes the token in milliseconds (default: 300000). Set to `0` to refresh only after an authentication retry.",

749 749 },

750Type / Values750 {

751 751 key: "model_providers.<id>.auth.cwd",

752`boolean`752 type: "string (path)",

753 753 description: "Working directory for the token command.",

754Details754 },

755 755 {

756Deprecated legacy toggle; prefer the top-level `web_search` setting.756 key: "model_providers.amazon-bedrock.aws.profile",

757 757 type: "string",

758Key758 description:

759 759 "AWS profile name used by the built-in `amazon-bedrock` provider.",

760`features.web_search_cached`760 },

761 761 {

762Type / Values762 key: "model_providers.amazon-bedrock.aws.region",

763 763 type: "string",

764`boolean`764 description: "AWS region used by the built-in `amazon-bedrock` provider.",

765 765 },

766Details766 {

767 767 key: "model_reasoning_effort",

768Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`.768 type: "minimal | low | medium | high | xhigh",

769 769 description:

770Key770 "Adjust reasoning effort for supported models (Responses API only; `xhigh` is model-dependent).",

771 771 },

772`features.web_search_request`772 {

773 773 key: "plan_mode_reasoning_effort",

774Type / Values774 type: "none | minimal | low | medium | high | xhigh",

775 775 description:

776`boolean`776 "Plan-mode-specific reasoning override. When unset, Plan mode uses its built-in preset default.",

777 777 },

778Details778 {

779 779 key: "model_reasoning_summary",

780Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`.780 type: "auto | concise | detailed | none",

781 781 description:

782Key782 "Select reasoning summary detail or disable summaries entirely.",

783 783 },

784`feedback.enabled`784 {

785 785 key: "model_verbosity",

786Type / Values786 type: "low | medium | high",

787 787 description:

788`boolean`788 "Optional GPT-5 Responses API verbosity override; when unset, the selected model/preset default is used.",

789 789 },

790Details790 {

791 791 key: "model_supports_reasoning_summaries",

792Enable feedback submission via `/feedback` across Codex surfaces (default: true).792 type: "boolean",

793 793 description: "Force Codex to send or not send reasoning metadata.",

794Key794 },

795 795 {

796`file_opener`796 key: "shell_environment_policy.inherit",

797 797 type: "all | core | none",

798Type / Values798 description:

799 799 "Baseline environment inheritance when spawning subprocesses.",

800`vscode | vscode-insiders | windsurf | cursor | none`800 },

801 801 {

802Details802 key: "shell_environment_policy.ignore_default_excludes",

803 803 type: "boolean",

804URI scheme used to open citations from Codex output (default: `vscode`).804 description:

805 805 "Keep variables containing KEY/SECRET/TOKEN before other filters run.",

806Key806 },

807 807 {

808`forced_chatgpt_workspace_id`808 key: "shell_environment_policy.exclude",

809 809 type: "array<string>",

810Type / Values810 description:

811 811 "Glob patterns for removing environment variables after the defaults.",

812`string (uuid)`812 },

813 813 {

814Details814 key: "shell_environment_policy.include_only",

815 815 type: "array<string>",

816Limit ChatGPT logins to a specific workspace identifier.816 description:

817 817 "Whitelist of patterns; when set only matching variables are kept.",

818Key818 },

819 819 {

820`forced_login_method`820 key: "shell_environment_policy.set",

821 821 type: "map<string,string>",

822Type / Values822 description:

823 823 "Explicit environment overrides injected into every subprocess.",

824`chatgpt | api`824 },

825 825 {

826Details826 key: "shell_environment_policy.experimental_use_profile",

827 827 type: "boolean",

828Restrict Codex to a specific authentication method.828 description: "Use the user shell profile when spawning subprocesses.",