config-reference.md Codex Docs, 2026-03-05 18:41 UTC → 2026-05-18 22:01 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).9Project-scoped config can't override machine-local provider, auth,

10 10notification, profile, or telemetry routing keys. Codex ignores

11| Key | Type / Values | Details |11`openai_base_url`, `chatgpt_base_url`, `model_provider`, `model_providers`,

12| --- | --- | --- |12`notify`, `profile`, `profiles`, `experimental_realtime_ws_base_url`, and

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`otel` when they appear in a project-local `.codex/config.toml`; put those in

14| `agents.<name>.description` | `string` | Role guidance shown to Codex when choosing and spawning that agent type. |14user-level config instead.

15| `agents.job_max_runtime_seconds` | `number` | Default per-worker timeout for `spawn_agents_on_csv` jobs. When unset, the tool falls back to 1800 seconds per worker. |15 

16| `agents.max_depth` | `number` | Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1). |16For 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).

17| `agents.max_threads` | `number` | Maximum number of agent threads that can be open concurrently. |17 

18| `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. |18<ConfigTable

19| `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. |19 options={[

20| `approval_policy.reject.mcp_elicitations` | `boolean` | When `true`, MCP elicitation prompts are auto-rejected instead of shown to the user. |20 {

21| `approval_policy.reject.rules` | `boolean` | When `true`, approvals triggered by execpolicy `prompt` rules are auto-rejected. |21 key: "model",

22| `approval_policy.reject.sandbox_approval` | `boolean` | When `true`, sandbox escalation approval prompts are auto-rejected. |22 type: "string",

23| `apps._default.destructive_enabled` | `boolean` | Default allow/deny for app tools with `destructive_hint = true`. |23 description: "Model to use (e.g., `gpt-5.5`).",

24| `apps._default.enabled` | `boolean` | Default app enabled state for all apps unless overridden per app. |24 },

25| `apps._default.open_world_enabled` | `boolean` | Default allow/deny for app tools with `open_world_hint = true`. |25 {

26| `apps.<id>.default_tools_approval_mode` | `auto | prompt | approve` | Default approval behavior for tools in this app unless a per-tool override exists. |26 key: "review_model",

27| `apps.<id>.default_tools_enabled` | `boolean` | Default enabled state for tools in this app unless a per-tool override exists. |27 type: "string",

28| `apps.<id>.destructive_enabled` | `boolean` | Allow or block tools in this app that advertise `destructive_hint = true`. |28 description:

29| `apps.<id>.enabled` | `boolean` | Enable or disable a specific app/connector by id (default: true). |29 "Optional model override used by `/review` (defaults to the current session model).",

30| `apps.<id>.open_world_enabled` | `boolean` | Allow or block tools in this app that advertise `open_world_hint = true`. |30 },

31| `apps.<id>.tools.<tool>.approval_mode` | `auto | prompt | approve` | Per-tool approval behavior override for a single app tool. |31 {

32| `apps.<id>.tools.<tool>.enabled` | `boolean` | Per-tool enabled override for an app tool (for example `repos/list`). |32 key: "model_provider",

33| `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. |33 type: "string",

34| `chatgpt_base_url` | `string` | Override the base URL used during the ChatGPT login flow. |34 description: "Provider id from `model_providers` (default: `openai`).",

35| `check_for_update_on_startup` | `boolean` | Check for Codex updates on startup (set to false only when updates are centrally managed). |35 },

36| `cli_auth_credentials_store` | `file | keyring | auto` | Control where the CLI stores cached credentials (file-based auth.json vs OS keychain). |36 {

37| `compact_prompt` | `string` | Inline override for the history compaction prompt. |37 key: "openai_base_url",

38| `developer_instructions` | `string` | Additional developer instructions injected into the session (optional). |38 type: "string",

39| `disable_paste_burst` | `boolean` | Disable burst-paste detection in the TUI. |39 description:

40| `experimental_compact_prompt_file` | `string (path)` | Load the compaction prompt override from a file (experimental). |40 "Base URL override for the built-in `openai` model provider.",

41| `experimental_use_freeform_apply_patch` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform` or `codex --enable apply_patch_freeform`. |41 },

42| `experimental_use_unified_exec_tool` | `boolean` | Legacy name for enabling unified exec; prefer `[features].unified_exec` or `codex --enable unified_exec`. |42 {

43| `features.apply_patch_freeform` | `boolean` | Expose the freeform `apply_patch` tool (experimental). |43 key: "model_context_window",

44| `features.apps` | `boolean` | Enable ChatGPT Apps/connectors support (experimental). |44 type: "number",

45| `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). |45 description: "Context window tokens available to the active model.",

46| `features.child_agents_md` | `boolean` | Append AGENTS.md scope/precedence guidance even when no AGENTS.md is present (experimental). |46 },

47| `features.collaboration_modes` | `boolean` | Enable collaboration modes such as plan mode (stable; on by default). |47 {

48| `features.multi_agent` | `boolean` | Enable multi-agent collaboration tools (`spawn_agent`, `send_input`, `resume_agent`, `wait`, `close_agent`, and `spawn_agents_on_csv`) (experimental; off by default). |48 key: "model_auto_compact_token_limit",

49| `features.personality` | `boolean` | Enable personality selection controls (stable; on by default). |49 type: "number",

50| `features.powershell_utf8` | `boolean` | Force PowerShell UTF-8 output (defaults to true). |50 description:

51| `features.remote_models` | `boolean` | Refresh remote model list before showing readiness (experimental). |51 "Token threshold that triggers automatic history compaction (unset uses model defaults).",

52| `features.request_rule` | `boolean` | Enable Smart approvals (`prefix_rule` suggestions on escalation requests; stable; on by default). |52 },

53| `features.runtime_metrics` | `boolean` | Show runtime metrics summary in TUI turn separators (experimental). |53 {

54| `features.search_tool` | `boolean` | Enable `search_tool_bm25` for Apps tool discovery before invoking app MCP tools (experimental). |54 key: "model_catalog_json",

55| `features.shell_snapshot` | `boolean` | Snapshot shell environment to speed up repeated commands (beta). |55 type: "string (path)",

56| `features.shell_tool` | `boolean` | Enable the default `shell` tool for running commands (stable; on by default). |56 description:

57| `features.unified_exec` | `boolean` | Use the unified PTY-backed exec tool (beta). |57 "Optional path to a JSON model catalog loaded on startup. Profile-level `profiles.<name>.model_catalog_json` can override this per profile.",

58| `features.use_linux_sandbox_bwrap` | `boolean` | Use the bubblewrap-based Linux sandbox pipeline (experimental; off by default). |58 },

59| `features.web_search` | `boolean` | Deprecated legacy toggle; prefer the top-level `web_search` setting. |59 {

60| `features.web_search_cached` | `boolean` | Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`. |60 key: "oss_provider",

61| `features.web_search_request` | `boolean` | Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`. |61 type: "lmstudio | ollama",

62| `feedback.enabled` | `boolean` | Enable feedback submission via `/feedback` across Codex surfaces (default: true). |62 description:

63| `file_opener` | `vscode | vscode-insiders | windsurf | cursor | none` | URI scheme used to open citations from Codex output (default: `vscode`). |63 "Default local provider used when running with `--oss` (defaults to prompting if unset).",

64| `forced_chatgpt_workspace_id` | `string (uuid)` | Limit ChatGPT logins to a specific workspace identifier. |64 },

65| `forced_login_method` | `chatgpt | api` | Restrict Codex to a specific authentication method. |65 {

66| `hide_agent_reasoning` | `boolean` | Suppress reasoning events in both the TUI and `codex exec` output. |66 key: "approval_policy",

67| `history.max_bytes` | `number` | If set, caps the history file size in bytes by dropping oldest entries. |67 type: "untrusted | on-request | never | { granular = { sandbox_approval = bool, rules = bool, mcp_elicitations = bool, request_permissions = bool, skill_approval = bool } }",

68| `history.persistence` | `save-all | none` | Control whether Codex saves session transcripts to history.jsonl. |68 description:

69| `include_apply_patch_tool` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |69 "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.",

70| `instructions` | `string` | Reserved for future use; prefer `model_instructions_file` or `AGENTS.md`. |70 },

71| `log_dir` | `string (path)` | Directory where Codex writes log files (for example `codex-tui.log`); defaults to `$CODEX_HOME/log`. |71 {

72| `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. |72 key: "approval_policy.granular.sandbox_approval",

73| `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. |73 type: "boolean",

74| `mcp_oauth_credentials_store` | `auto | file | keyring` | Preferred store for MCP OAuth credentials. |74 description:

75| `mcp_servers.<id>.args` | `array<string>` | Arguments passed to the MCP stdio server command. |75 "When `true`, sandbox escalation approval prompts are allowed to surface.",

76| `mcp_servers.<id>.bearer_token_env_var` | `string` | Environment variable sourcing the bearer token for an MCP HTTP server. |76 },

77| `mcp_servers.<id>.command` | `string` | Launcher command for an MCP stdio server. |77 {

78| `mcp_servers.<id>.cwd` | `string` | Working directory for the MCP stdio server process. |78 key: "approval_policy.granular.rules",

79| `mcp_servers.<id>.disabled_tools` | `array<string>` | Deny list applied after `enabled_tools` for the MCP server. |79 type: "boolean",

80| `mcp_servers.<id>.enabled` | `boolean` | Disable an MCP server without removing its configuration. |80 description:

81| `mcp_servers.<id>.enabled_tools` | `array<string>` | Allow list of tool names exposed by the MCP server. |81 "When `true`, approvals triggered by execpolicy `prompt` rules are allowed to surface.",

82| `mcp_servers.<id>.env` | `map<string,string>` | Environment variables forwarded to the MCP stdio server. |82 },

83| `mcp_servers.<id>.env_http_headers` | `map<string,string>` | HTTP headers populated from environment variables for an MCP HTTP server. |83 {

84| `mcp_servers.<id>.env_vars` | `array<string>` | Additional environment variables to whitelist for an MCP stdio server. |84 key: "approval_policy.granular.mcp_elicitations",

85| `mcp_servers.<id>.http_headers` | `map<string,string>` | Static HTTP headers included with each MCP HTTP request. |85 type: "boolean",

86| `mcp_servers.<id>.required` | `boolean` | When true, fail startup/resume if this enabled MCP server cannot initialize. |86 description:

87| `mcp_servers.<id>.startup_timeout_ms` | `number` | Alias for `startup_timeout_sec` in milliseconds. |87 "When `true`, MCP elicitation prompts are allowed to surface instead of being auto-rejected.",

88| `mcp_servers.<id>.startup_timeout_sec` | `number` | Override the default 10s startup timeout for an MCP server. |88 },

89| `mcp_servers.<id>.tool_timeout_sec` | `number` | Override the default 60s per-tool timeout for an MCP server. |89 {

90| `mcp_servers.<id>.url` | `string` | Endpoint for an MCP streamable HTTP server. |90 key: "approval_policy.granular.request_permissions",

91| `model` | `string` | Model to use (e.g., `gpt-5-codex`). |91 type: "boolean",

92| `model_auto_compact_token_limit` | `number` | Token threshold that triggers automatic history compaction (unset uses model defaults). |92 description:

93| `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. |93 "When `true`, prompts from the `request_permissions` tool are allowed to surface.",

94| `model_context_window` | `number` | Context window tokens available to the active model. |94 },

95| `model_instructions_file` | `string (path)` | Replacement for built-in instructions instead of `AGENTS.md`. |95 {

96| `model_provider` | `string` | Provider id from `model_providers` (default: `openai`). |96 key: "approval_policy.granular.skill_approval",

97| `model_providers.<id>.base_url` | `string` | API base URL for the model provider. |97 type: "boolean",

98| `model_providers.<id>.env_http_headers` | `map<string,string>` | HTTP headers populated from environment variables when present. |98 description:

99| `model_providers.<id>.env_key` | `string` | Environment variable supplying the provider API key. |99 "When `true`, skill-script approval prompts are allowed to surface.",

100| `model_providers.<id>.env_key_instructions` | `string` | Optional setup guidance for the provider API key. |100 },

101| `model_providers.<id>.experimental_bearer_token` | `string` | Direct bearer token for the provider (discouraged; use `env_key`). |101 {

102| `model_providers.<id>.http_headers` | `map<string,string>` | Static HTTP headers added to provider requests. |102 key: "approvals_reviewer",

103| `model_providers.<id>.name` | `string` | Display name for a custom model provider. |103 type: "user | auto_review",

104| `model_providers.<id>.query_params` | `map<string,string>` | Extra query parameters appended to provider requests. |104 description:

105| `model_providers.<id>.request_max_retries` | `number` | Retry count for HTTP requests to the provider (default: 4). |105 "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.",

106| `model_providers.<id>.requires_openai_auth` | `boolean` | The provider uses OpenAI authentication (defaults to false). |106 },

107| `model_providers.<id>.stream_idle_timeout_ms` | `number` | Idle timeout for SSE streams in milliseconds (default: 300000). |107 {

108| `model_providers.<id>.stream_max_retries` | `number` | Retry count for SSE streaming interruptions (default: 5). |108 key: "auto_review.policy",

109| `model_providers.<id>.wire_api` | `chat | responses` | Protocol used by the provider (defaults to `chat` if omitted). |109 type: "string",

110| `model_reasoning_effort` | `minimal | low | medium | high | xhigh` | Adjust reasoning effort for supported models (Responses API only; `xhigh` is model-dependent). |110 description:

111| `model_reasoning_summary` | `auto | concise | detailed | none` | Select reasoning summary detail or disable summaries entirely. |111 "Local Markdown policy instructions for automatic review. Managed `guardian_policy_config` takes precedence. Blank values are ignored.",

112| `model_supports_reasoning_summaries` | `boolean` | Force Codex to send or not send reasoning metadata. |112 },

113| `model_verbosity` | `low | medium | high` | Control GPT-5 Responses API verbosity (defaults to `medium`). |113 {

114| `notice.hide_full_access_warning` | `boolean` | Track acknowledgement of the full access warning prompt. |114 key: "allow_login_shell",

115| `notice.hide_gpt-5.1-codex-max_migration_prompt` | `boolean` | Track acknowledgement of the gpt-5.1-codex-max migration prompt. |115 type: "boolean",

116| `notice.hide_gpt5_1_migration_prompt` | `boolean` | Track acknowledgement of the GPT-5.1 migration prompt. |116 description:

117| `notice.hide_rate_limit_model_nudge` | `boolean` | Track opt-out of the rate limit model switch reminder. |117 "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.",

118| `notice.hide_world_writable_warning` | `boolean` | Track acknowledgement of the Windows world-writable directories warning. |118 },

119| `notice.model_migrations` | `map<string,string>` | Track acknowledged model migrations as old->new mappings. |119 {

120| `notify` | `array<string>` | Command invoked for notifications; receives a JSON payload from Codex. |120 key: "sandbox_mode",

121| `oss_provider` | `lmstudio | ollama` | Default local provider used when running with `--oss` (defaults to prompting if unset). |121 type: "read-only | workspace-write | danger-full-access",

122| `otel.environment` | `string` | Environment tag applied to emitted OpenTelemetry events (default: `dev`). |122 description:

123| `otel.exporter` | `none | otlp-http | otlp-grpc` | Select the OpenTelemetry exporter and provide any endpoint metadata. |123 "Sandbox policy for filesystem and network access during command execution.",

124| `otel.exporter.<id>.endpoint` | `string` | Exporter endpoint for OTEL logs. |124 },

125| `otel.exporter.<id>.headers` | `map<string,string>` | Static headers included with OTEL exporter requests. |125 {

126| `otel.exporter.<id>.protocol` | `binary | json` | Protocol used by the OTLP/HTTP exporter. |126 key: "sandbox_workspace_write.writable_roots",

127| `otel.exporter.<id>.tls.ca-certificate` | `string` | CA certificate path for OTEL exporter TLS. |127 type: "array<string>",

128| `otel.exporter.<id>.tls.client-certificate` | `string` | Client certificate path for OTEL exporter TLS. |128 description:

129| `otel.exporter.<id>.tls.client-private-key` | `string` | Client private key path for OTEL exporter TLS. |129 'Additional writable roots when `sandbox_mode = "workspace-write"`.',

130| `otel.log_user_prompt` | `boolean` | Opt in to exporting raw user prompts with OpenTelemetry logs. |130 },

131| `otel.trace_exporter` | `none | otlp-http | otlp-grpc` | Select the OpenTelemetry trace exporter and provide any endpoint metadata. |131 {

132| `otel.trace_exporter.<id>.endpoint` | `string` | Trace exporter endpoint for OTEL logs. |132 key: "sandbox_workspace_write.network_access",

133| `otel.trace_exporter.<id>.headers` | `map<string,string>` | Static headers included with OTEL trace exporter requests. |133 type: "boolean",

134| `otel.trace_exporter.<id>.protocol` | `binary | json` | Protocol used by the OTLP/HTTP trace exporter. |134 description:

135| `otel.trace_exporter.<id>.tls.ca-certificate` | `string` | CA certificate path for OTEL trace exporter TLS. |135 "Allow outbound network access inside the workspace-write sandbox.",

136| `otel.trace_exporter.<id>.tls.client-certificate` | `string` | Client certificate path for OTEL trace exporter TLS. |136 },

137| `otel.trace_exporter.<id>.tls.client-private-key` | `string` | Client private key path for OTEL trace exporter TLS. |137 {

138| `personality` | `none | friendly | pragmatic` | Default communication style for models that advertise `supportsPersonality`; can be overridden per thread/turn or via `/personality`. |138 key: "sandbox_workspace_write.exclude_tmpdir_env_var",

139| `profile` | `string` | Default profile applied at startup (equivalent to `--profile`). |139 type: "boolean",

140| `profiles.<name>.*` | `various` | Profile-scoped overrides for any of the supported configuration keys. |140 description:

141| `profiles.<name>.experimental_use_freeform_apply_patch` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |141 "Exclude `$TMPDIR` from writable roots in workspace-write mode.",

142| `profiles.<name>.experimental_use_unified_exec_tool` | `boolean` | Legacy name for enabling unified exec; prefer `[features].unified_exec`. |142 },

143| `profiles.<name>.include_apply_patch_tool` | `boolean` | Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform`. |143 {

144| `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). |144 key: "sandbox_workspace_write.exclude_slash_tmp",

145| `profiles.<name>.oss_provider` | `lmstudio | ollama` | Profile-scoped OSS provider for `--oss` sessions. |145 type: "boolean",

146| `profiles.<name>.personality` | `none | friendly | pragmatic` | Profile-scoped communication style override for supported models. |146 description:

147| `profiles.<name>.web_search` | `disabled | cached | live` | Profile-scoped web search mode override (default: `"cached"`). |147 "Exclude `/tmp` from writable roots in workspace-write mode.",

148| `project_doc_fallback_filenames` | `array<string>` | Additional filenames to try when `AGENTS.md` is missing. |148 },

149| `project_doc_max_bytes` | `number` | Maximum bytes read from `AGENTS.md` when building project instructions. |149 {

150| `project_root_markers` | `array<string>` | List of project root marker filenames; used when searching parent directories for the project root. |150 key: "windows.sandbox",

151| `projects.<path>.trust_level` | `string` | Mark a project or worktree as trusted or untrusted (`"trusted"` | `"untrusted"`). Untrusted projects skip project-scoped `.codex/` layers. |151 type: "unelevated | elevated",

152| `review_model` | `string` | Optional model override used by `/review` (defaults to the current session model). |152 description:

153| `sandbox_mode` | `read-only | workspace-write | danger-full-access` | Sandbox policy for filesystem and network access during command execution. |153 "Windows-only native sandbox mode when running Codex natively on Windows.",

154| `sandbox_workspace_write.exclude_slash_tmp` | `boolean` | Exclude `/tmp` from writable roots in workspace-write mode. |154 },

155| `sandbox_workspace_write.exclude_tmpdir_env_var` | `boolean` | Exclude `$TMPDIR` from writable roots in workspace-write mode. |155 {

156| `sandbox_workspace_write.network_access` | `boolean` | Allow outbound network access inside the workspace-write sandbox. |156 key: "windows.sandbox_private_desktop",

157| `sandbox_workspace_write.writable_roots` | `array<string>` | Additional writable roots when `sandbox_mode = "workspace-write"`. |157 type: "boolean",

158| `shell_environment_policy.exclude` | `array<string>` | Glob patterns for removing environment variables after the defaults. |158 description:

159| `shell_environment_policy.experimental_use_profile` | `boolean` | Use the user shell profile when spawning subprocesses. |159 "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.",

160| `shell_environment_policy.ignore_default_excludes` | `boolean` | Keep variables containing KEY/SECRET/TOKEN before other filters run. |160 },

161| `shell_environment_policy.include_only` | `array<string>` | Whitelist of patterns; when set only matching variables are kept. |161 {

162| `shell_environment_policy.inherit` | `all | core | none` | Baseline environment inheritance when spawning subprocesses. |162 key: "notify",

163| `shell_environment_policy.set` | `map<string,string>` | Explicit environment overrides injected into every subprocess. |163 type: "array<string>",

164| `show_raw_agent_reasoning` | `boolean` | Surface raw reasoning content when the active model emits it. |164 description:

165| `skills.config` | `array<object>` | Per-skill enablement overrides stored in config.toml. |165 "Command invoked for notifications; receives a JSON payload from Codex.",

166| `skills.config.<index>.enabled` | `boolean` | Enable or disable the referenced skill. |166 },

167| `skills.config.<index>.path` | `string (path)` | Path to a skill folder containing `SKILL.md`. |167 {

168| `sqlite_home` | `string (path)` | Directory where Codex stores the SQLite-backed state DB used by agent jobs and other resumable runtime state. |168 key: "check_for_update_on_startup",

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

170| `tool_output_token_limit` | `number` | Token budget for storing individual tool/function outputs in history. |170 description:

171| `tools.web_search` | `boolean` | Deprecated legacy toggle for web search; prefer the top-level `web_search` setting. |171 "Check for Codex updates on startup (set to false only when updates are centrally managed).",

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 {

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

175| `tui.notification_method` | `auto | osc9 | bel` | Notification method for unfocused terminal notifications (default: auto). |175 type: "boolean",

176| `tui.notifications` | `boolean | array<string>` | Enable TUI notifications; optionally restrict to specific event types. |176 description:

177| `tui.show_tooltips` | `boolean` | Show onboarding tooltips in the TUI welcome screen (default: true). |177 "Enable feedback submission via `/feedback` across Codex surfaces (default: true).",

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 {

180| `windows_wsl_setup_acknowledged` | `boolean` | Track Windows onboarding acknowledgement (Windows only). |180 key: "analytics.enabled",

181| `windows.sandbox` | `unelevated | elevated` | Windows-only native sandbox mode when running Codex natively on Windows. |181 type: "boolean",

182 182 description:

183Key183 "Enable or disable analytics for this machine/profile. When unset, the client default applies.",

184 184 },

185`agents.<name>.config_file`185 {

186 186 key: "instructions",

187Type / Values187 type: "string",

188 188 description:

189`string (path)`189 "Reserved for future use; prefer `model_instructions_file` or `AGENTS.md`.",

190 190 },

191Details191 {

192 192 key: "developer_instructions",

193Path to a TOML config layer for that role; relative paths resolve from the config file that declares the role.193 type: "string",

194 194 description:

195Key195 "Additional developer instructions injected into the session (optional).",

196 196 },

197`agents.<name>.description`197 {

198 198 key: "log_dir",

199Type / Values199 type: "string (path)",

200 200 description:

201`string`201 "Directory where Codex writes log files (for example `codex-tui.log`); defaults to `$CODEX_HOME/log`.",

202 202 },

203Details203 {

204 204 key: "sqlite_home",

205Role guidance shown to Codex when choosing and spawning that agent type.205 type: "string (path)",

206 206 description:

207Key207 "Directory where Codex stores the SQLite-backed state DB used by agent jobs and other resumable runtime state.",

208 208 },

209`agents.job_max_runtime_seconds`209 {

210 210 key: "compact_prompt",

211Type / Values211 type: "string",

212 212 description: "Inline override for the history compaction prompt.",

213`number`213 },

214 214 {

215Details215 key: "commit_attribution",

216 216 type: "string",

217Default per-worker timeout for `spawn_agents_on_csv` jobs. When unset, the tool falls back to 1800 seconds per worker.217 description:

218 218 'Commit co-author trailer used when `[features].codex_git_commit` is enabled. Defaults to `Codex <noreply@openai.com>`; set `""` to disable.',

219Key219 },

220 220 {

221`agents.max_depth`221 key: "model_instructions_file",

222 222 type: "string (path)",

223Type / Values223 description:

224 224 "Replacement for built-in instructions instead of `AGENTS.md`.",

225`number`225 },

226 226 {

227Details227 key: "personality",

228 228 type: "none | friendly | pragmatic",

229Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1).229 description:

230 230 "Default communication style for models that advertise `supportsPersonality`; can be overridden per thread/turn or via `/personality`.",

231Key231 },

232 232 {

233`agents.max_threads`233 key: "service_tier",

234 234 type: "string",

235Type / Values235 description:

236 236 "Preferred service tier for new turns. Built-in values include `flex` and `fast`; legacy `fast` config maps to the request value `priority`, and catalog-provided tier IDs can also be stored.",

237`number`237 },

238 238 {

239Details239 key: "experimental_compact_prompt_file",

240 240 type: "string (path)",

241Maximum number of agent threads that can be open concurrently.241 description:

242 242 "Load the compaction prompt override from a file (experimental).",

243Key243 },

244 244 {

245`allow_login_shell`245 key: "skills.config",

246 246 type: "array<object>",

247Type / Values247 description: "Per-skill enablement overrides stored in config.toml.",

248 248 },

249`boolean`249 {

250 250 key: "skills.config.<index>.path",

251Details251 type: "string (path)",

252 252 description: "Path to a skill folder containing `SKILL.md`.",

253Allow 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.253 },

254 254 {

255Key255 key: "skills.config.<index>.enabled",

256 256 type: "boolean",

257`approval_policy`257 description: "Enable or disable the referenced skill.",

258 258 },

259Type / Values259 {

260 260 key: "apps.<id>.enabled",

261`untrusted | on-request | never | { reject = { sandbox_approval = bool, rules = bool, mcp_elicitations = bool } }`261 type: "boolean",

262 262 description:

263Details263 "Enable or disable a specific app/connector by id (default: true).",

264 264 },

265Controls 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.265 {

266 266 key: "apps._default.enabled",

267Key267 type: "boolean",

268 268 description:

269`approval_policy.reject.mcp_elicitations`269 "Default app enabled state for all apps unless overridden per app.",

270 270 },

271Type / Values271 {

272 272 key: "apps._default.destructive_enabled",

273`boolean`273 type: "boolean",

274 274 description:

275Details275 "Default allow/deny for app tools with `destructive_hint = true`.",

276 276 },

277When `true`, MCP elicitation prompts are auto-rejected instead of shown to the user.277 {

278 278 key: "apps._default.open_world_enabled",

279Key279 type: "boolean",

280 280 description:

281`approval_policy.reject.rules`281 "Default allow/deny for app tools with `open_world_hint = true`.",

282 282 },

283Type / Values283 {

284 284 key: "apps.<id>.destructive_enabled",

285`boolean`285 type: "boolean",

286 286 description:

287Details287 "Allow or block tools in this app that advertise `destructive_hint = true`.",

288 288 },

289When `true`, approvals triggered by execpolicy `prompt` rules are auto-rejected.289 {

290 290 key: "apps.<id>.open_world_enabled",

291Key291 type: "boolean",

292 292 description:

293`approval_policy.reject.sandbox_approval`293 "Allow or block tools in this app that advertise `open_world_hint = true`.",

294 294 },

295Type / Values295 {

296 296 key: "apps.<id>.default_tools_enabled",

297`boolean`297 type: "boolean",

298 298 description:

299Details299 "Default enabled state for tools in this app unless a per-tool override exists.",

300 300 },

301When `true`, sandbox escalation approval prompts are auto-rejected.301 {

302 302 key: "apps.<id>.default_tools_approval_mode",

303Key303 type: "auto | prompt | approve",

304 304 description:

305`apps._default.destructive_enabled`305 "Default approval behavior for tools in this app unless a per-tool override exists.",

306 306 },

307Type / Values307 {

308 308 key: "apps.<id>.tools.<tool>.enabled",

309`boolean`309 type: "boolean",

310 310 description:

311Details311 "Per-tool enabled override for an app tool (for example `repos/list`).",

312 312 },

313Default allow/deny for app tools with `destructive_hint = true`.313 {

314 314 key: "apps.<id>.tools.<tool>.approval_mode",

315Key315 type: "auto | prompt | approve",

316 316 description: "Per-tool approval behavior override for a single app tool.",

317`apps._default.enabled`317 },

318 318 {

319Type / Values319 key: "tool_suggest.discoverables",

320 320 type: "array<table>",

321`boolean`321 description:

322 322 'Allow tool suggestions for additional discoverable connectors or plugins. Each entry uses `type = "connector"` or `"plugin"` and an `id`.',

323Details323 },

324 324 {

325Default app enabled state for all apps unless overridden per app.325 key: "tool_suggest.disabled_tools",

326 326 type: "array<table>",

327Key327 description:

328 328 'Disable suggestions for specific discoverable connectors or plugins. Each entry uses `type = "connector"` or `"plugin"` and an `id`.',

329`apps._default.open_world_enabled`329 },

330 330 {

331Type / Values331 key: "features.apps",

332 332 type: "boolean",

333`boolean`333 description: "Enable ChatGPT Apps/connectors support (experimental).",

334 334 },

335Details335 {

336 336 key: "features.hooks",

337Default allow/deny for app tools with `open_world_hint = true`.337 type: "boolean",

338 338 description:

339Key339 "Enable lifecycle hooks loaded from `hooks.json` or inline `[hooks]` config. `features.codex_hooks` is a deprecated alias.",

340 340 },

341`apps.<id>.default_tools_approval_mode`341 {

342 342 key: "features.codex_git_commit",

343Type / Values343 type: "boolean",

344 344 description:

345`auto | prompt | approve`345 "Enable Codex-generated git commits. When enabled, Codex uses `commit_attribution` to append a `Co-authored-by:` trailer to generated commit messages.",

346 346 },

347Details347 {

348 348 key: "hooks",

349Default approval behavior for tools in this app unless a per-tool override exists.349 type: "table",

350 350 description:

351Key351 "Lifecycle hooks configured inline in `config.toml`. Uses the same event schema as `hooks.json`; see the Hooks guide for examples and supported events.",

352 352 },

353`apps.<id>.default_tools_enabled`353 {

354 354 key: "features.plugin_hooks",

355Type / Values355 type: "boolean",

356 356 description:

357`boolean`357 "Opt into lifecycle hooks bundled with enabled plugins. Off by default in this release; set to `true` to opt in.",

358 358 },

359Details359 {

360 360 key: "features.memories",

361Default enabled state for tools in this app unless a per-tool override exists.361 type: "boolean",

362 362 description: "Enable [Memories](https://developers.openai.com/codex/memories) (off by default).",

363Key363 },

364 364 {

365`apps.<id>.destructive_enabled`365 key: "mcp_servers.<id>.command",

366 366 type: "string",

367Type / Values367 description: "Launcher command for an MCP stdio server.",

368 368 },

369`boolean`369 {

370 370 key: "mcp_servers.<id>.args",

371Details371 type: "array<string>",

372 372 description: "Arguments passed to the MCP stdio server command.",

373Allow or block tools in this app that advertise `destructive_hint = true`.373 },

374 374 {

375Key375 key: "mcp_servers.<id>.env",

376 376 type: "map<string,string>",

377`apps.<id>.enabled`377 description: "Environment variables forwarded to the MCP stdio server.",

378 378 },

379Type / Values379 {

380 380 key: "mcp_servers.<id>.env_vars",

381`boolean`381 type: 'array<string | { name = string, source = "local" | "remote" }>',

382 382 description:

383Details383 '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.',

384 384 },

385Enable or disable a specific app/connector by id (default: true).385 {

386 386 key: "mcp_servers.<id>.cwd",

387Key387 type: "string",

388 388 description: "Working directory for the MCP stdio server process.",

389`apps.<id>.open_world_enabled`389 },

390 390 {

391Type / Values391 key: "mcp_servers.<id>.url",

392 392 type: "string",

393`boolean`393 description: "Endpoint for an MCP streamable HTTP server.",

394 394 },

395Details395 {

396 396 key: "mcp_servers.<id>.bearer_token_env_var",

397Allow or block tools in this app that advertise `open_world_hint = true`.397 type: "string",

398 398 description:

399Key399 "Environment variable sourcing the bearer token for an MCP HTTP server.",

400 400 },

401`apps.<id>.tools.<tool>.approval_mode`401 {

402 402 key: "mcp_servers.<id>.http_headers",

403Type / Values403 type: "map<string,string>",

404 404 description: "Static HTTP headers included with each MCP HTTP request.",

405`auto | prompt | approve`405 },

406 406 {

407Details407 key: "mcp_servers.<id>.env_http_headers",

408 408 type: "map<string,string>",

409Per-tool approval behavior override for a single app tool.409 description:

410 410 "HTTP headers populated from environment variables for an MCP HTTP server.",

411Key411 },

412 412 {

413`apps.<id>.tools.<tool>.enabled`413 key: "mcp_servers.<id>.enabled",

414 414 type: "boolean",

415Type / Values415 description: "Disable an MCP server without removing its configuration.",

416 416 },

417`boolean`417 {

418 418 key: "mcp_servers.<id>.required",

419Details419 type: "boolean",

420 420 description:

421Per-tool enabled override for an app tool (for example `repos/list`).421 "When true, fail startup/resume if this enabled MCP server cannot initialize.",

422 422 },

423Key423 {

424 424 key: "mcp_servers.<id>.startup_timeout_sec",

425`background_terminal_max_timeout`425 type: "number",

426 426 description:

427Type / Values427 "Override the default 10s startup timeout for an MCP server.",

428 428 },

429`number`429 {

430 430 key: "mcp_servers.<id>.startup_timeout_ms",

431Details431 type: "number",

432 432 description: "Alias for `startup_timeout_sec` in milliseconds.",

433Maximum poll window in milliseconds for empty `write_stdin` polls (background terminal polling). Default: `300000` (5 minutes). Replaces the older `background_terminal_timeout` key.433 },

434 434 {

435Key435 key: "mcp_servers.<id>.tool_timeout_sec",

436 436 type: "number",

437`chatgpt_base_url`437 description:

438 438 "Override the default 60s per-tool timeout for an MCP server.",

439Type / Values439 },

440 440 {

441`string`441 key: "mcp_servers.<id>.enabled_tools",

442 442 type: "array<string>",

443Details443 description: "Allow list of tool names exposed by the MCP server.",

444 444 },

445Override the base URL used during the ChatGPT login flow.445 {

446 446 key: "mcp_servers.<id>.disabled_tools",

447Key447 type: "array<string>",

448 448 description:

449`check_for_update_on_startup`449 "Deny list applied after `enabled_tools` for the MCP server.",

450 450 },

451Type / Values451 {

452 452 key: "mcp_servers.<id>.default_tools_approval_mode",

453`boolean`453 type: "auto | prompt | approve",

454 454 description:

455Details455 "Default approval behavior for MCP tools on this server unless a per-tool override exists.",

456 456 },

457Check for Codex updates on startup (set to false only when updates are centrally managed).457 {

458 458 key: "mcp_servers.<id>.tools.<tool>.approval_mode",

459Key459 type: "auto | prompt | approve",

460 460 description:

461`cli_auth_credentials_store`461 "Per-tool approval behavior override for one MCP tool on this server.",

462 462 },

463Type / Values463 {

464 464 key: "mcp_servers.<id>.scopes",

465`file | keyring | auto`465 type: "array<string>",

466 466 description:

467Details467 "OAuth scopes to request when authenticating to that MCP server.",

468 468 },

469Control where the CLI stores cached credentials (file-based auth.json vs OS keychain).469 {

470 470 key: "mcp_servers.<id>.oauth_resource",

471Key471 type: "string",

472 472 description:

473`compact_prompt`473 "Optional RFC 8707 OAuth resource parameter to include during MCP login.",

474 474 },

475Type / Values475 {

476 476 key: "mcp_servers.<id>.experimental_environment",

477`string`477 type: "local | remote",

478 478 description:

479Details479 "Experimental placement for an MCP server. `remote` starts stdio servers through a remote executor environment; streamable HTTP remote placement is not implemented.",

480 480 },

481Inline override for the history compaction prompt.481 {

482 482 key: "agents.max_threads",

483Key483 type: "number",

484 484 description:

485`developer_instructions`485 "Maximum number of agent threads that can be open concurrently. Defaults to `6` when unset.",

486 486 },

487Type / Values487 {

488 488 key: "agents.max_depth",

489`string`489 type: "number",

490 490 description:

491Details491 "Maximum nesting depth allowed for spawned agent threads (root sessions start at depth 0; default: 1).",

492 492 },

493Additional developer instructions injected into the session (optional).493 {

494 494 key: "agents.job_max_runtime_seconds",

495Key495 type: "number",

496 496 description:

497`disable_paste_burst`497 "Default per-worker timeout for `spawn_agents_on_csv` jobs. When unset, the tool falls back to 1800 seconds per worker.",

498 498 },

499Type / Values499 {

500 500 key: "agents.<name>.description",

501`boolean`501 type: "string",

502 502 description:

503Details503 "Role guidance shown to Codex when choosing and spawning that agent type.",

504 504 },

505Disable burst-paste detection in the TUI.505 {

506 506 key: "agents.<name>.config_file",

507Key507 type: "string (path)",

508 508 description:

509`experimental_compact_prompt_file`509 "Path to a TOML config layer for that role; relative paths resolve from the config file that declares the role.",

510 510 },

511Type / Values511 {

512 512 key: "agents.<name>.nickname_candidates",

513`string (path)`513 type: "array<string>",

514 514 description:

515Details515 "Optional pool of display nicknames for spawned agents in that role.",

516 516 },

517Load the compaction prompt override from a file (experimental).517 {

518 518 key: "memories.generate_memories",

519Key519 type: "boolean",

520 520 description:

521`experimental_use_freeform_apply_patch`521 "When `false`, newly created threads are not stored as memory-generation inputs. Defaults to `true`.",

522 522 },

523Type / Values523 {

524 524 key: "memories.use_memories",

525`boolean`525 type: "boolean",

526 526 description:

527Details527 "When `false`, Codex skips injecting existing memories into future sessions. Defaults to `true`.",

528 528 },

529Legacy name for enabling freeform apply\_patch; prefer `[features].apply_patch_freeform` or `codex --enable apply_patch_freeform`.529 {

530 530 key: "memories.disable_on_external_context",

531Key531 type: "boolean",

532 532 description:

533`experimental_use_unified_exec_tool`533 "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`.",

534 534 },

535Type / Values535 {

536 536 key: "memories.max_raw_memories_for_consolidation",

537`boolean`537 type: "number",

538 538 description:

539Details539 "Maximum recent raw memories retained for global consolidation. Defaults to `256` and is capped at `4096`.",

540 540 },

541Legacy name for enabling unified exec; prefer `[features].unified_exec` or `codex --enable unified_exec`.541 {

542 542 key: "memories.max_unused_days",

543Key543 type: "number",

544 544 description:

545`features.apply_patch_freeform`545 "Maximum days since a memory was last used before it becomes ineligible for consolidation. Defaults to `30` and is clamped to `0`-`365`.",

546 546 },

547Type / Values547 {

548 548 key: "memories.max_rollout_age_days",

549`boolean`549 type: "number",

550 550 description:

551Details551 "Maximum age of threads considered for memory generation. Defaults to `30` and is clamped to `0`-`90`.",

552 552 },

553Expose the freeform `apply_patch` tool (experimental).553 {

554 554 key: "memories.max_rollouts_per_startup",

555Key555 type: "number",

556 556 description:

557`features.apps`557 "Maximum rollout candidates processed per startup pass. Defaults to `16` and is capped at `128`.",

558 558 },

559Type / Values559 {

560 560 key: "memories.min_rollout_idle_hours",

561`boolean`561 type: "number",

562 562 description:

563Details563 "Minimum idle time before a thread is considered for memory generation. Defaults to `6` and is clamped to `1`-`48`.",

564 564 },

565Enable ChatGPT Apps/connectors support (experimental).565 {

566 566 key: "memories.min_rate_limit_remaining_percent",

567Key567 type: "number",

568 568 description:

569`features.apps_mcp_gateway`569 "Minimum remaining percentage required in Codex rate-limit windows before memory generation starts. Defaults to `25` and is clamped to `0`-`100`.",

570 570 },

571Type / Values571 {

572 572 key: "memories.extract_model",

573`boolean`573 type: "string",

574 574 description: "Optional model override for per-thread memory extraction.",

575Details575 },

576 576 {

577Route Apps MCP calls through the OpenAI connectors MCP gateway (`https://api.openai.com/v1/connectors/mcp/`) instead of legacy routing (experimental).577 key: "memories.consolidation_model",

578 578 type: "string",

579Key579 description: "Optional model override for global memory consolidation.",

580 580 },

581`features.child_agents_md`581 {

582 582 key: "features.unified_exec",

583Type / Values583 type: "boolean",

584 584 description:

585`boolean`585 "Use the unified PTY-backed exec tool (stable; enabled by default except on Windows).",

586 586 },

587Details587 {

588 588 key: "features.shell_snapshot",

589Append AGENTS.md scope/precedence guidance even when no AGENTS.md is present (experimental).589 type: "boolean",

590 590 description:

591Key591 "Snapshot shell environment to speed up repeated commands (stable; on by default).",

592 592 },

593`features.collaboration_modes`593 {

594 594 key: "features.undo",

595Type / Values595 type: "boolean",

596 596 description: "Enable undo support (stable; off by default).",

597`boolean`597 },

598 598 {

599Details599 key: "features.multi_agent",

600 600 type: "boolean",

601Enable collaboration modes such as plan mode (stable; on by default).601 description:

602 602 "Enable multi-agent collaboration tools (`spawn_agent`, `send_input`, `resume_agent`, `wait_agent`, and `close_agent`) (stable; on by default).",

603Key603 },

604 604 {

605`features.multi_agent`605 key: "features.personality",

606 606 type: "boolean",

607Type / Values607 description:

608 608 "Enable personality selection controls (stable; on by default).",

609`boolean`609 },

610 610 {

611Details611 key: "features.network_proxy",

612 612 type: "boolean | table",

613Enable multi-agent collaboration tools (`spawn_agent`, `send_input`, `resume_agent`, `wait`, `close_agent`, and `spawn_agents_on_csv`) (experimental; off by default).613 description:

614 614 "Enable sandboxed networking. Use a table form when setting network policy options such as `domains` (experimental; off by default).",

615Key615 },

616 616 {

617`features.personality`617 key: "features.network_proxy.enabled",

618 618 type: "boolean",

619Type / Values619 description: "Enable sandboxed networking. Defaults to `false`.",

620 620 },

621`boolean`621 {

622 622 key: "features.network_proxy.domains",

623Details623 type: "map<string, allow | deny>",

624 624 description:

625Enable personality selection controls (stable; on by default).625 "Domain policy for sandboxed networking. Unset by default, which means no external destinations are allowed until you add `allow` rules. Supports exact hosts, `*.example.com` for subdomains only, `**.example.com` for apex plus subdomains, and global `*` allow rules; prefer scoped rules because `*` broadly opens public outbound access. Add `deny` rules for blocked destinations; `deny` wins on conflicts.",

626 626 },

627Key627 {

628 628 key: "features.network_proxy.unix_sockets",

629`features.powershell_utf8`629 type: "map<string, allow | none>",

630 630 description:

631Type / Values631 "Unix socket policy for sandboxed networking. Unset by default; add `allow` entries for permitted sockets.",

632 632 },

633`boolean`633 {

634 634 key: "features.network_proxy.allow_local_binding",

635Details635 type: "boolean",

636 636 description:

637Force PowerShell UTF-8 output (defaults to true).637 "Allow broader local/private-network access. Defaults to `false`; exact local IP literal or `localhost` allow rules can still permit specific local targets.",

638 638 },

639Key639 {

640 640 key: "features.network_proxy.enable_socks5",

641`features.remote_models`641 type: "boolean",

642 642 description: "Expose SOCKS5 support. Defaults to `true`.",

643Type / Values643 },

644 644 {

645`boolean`645 key: "features.network_proxy.enable_socks5_udp",

646 646 type: "boolean",

647Details647 description: "Allow UDP over SOCKS5. Defaults to `true`.",

648 648 },

649Refresh remote model list before showing readiness (experimental).649 {

650 650 key: "features.network_proxy.allow_upstream_proxy",

651Key651 type: "boolean",

652 652 description:

653`features.request_rule`653 "Allow chaining through an upstream proxy from the environment. Defaults to `true`.",

654 654 },

655Type / Values655 {

656 656 key: "features.network_proxy.dangerously_allow_non_loopback_proxy",

657`boolean`657 type: "boolean",

658 658 description:

659Details659 "Permit non-loopback listener addresses. Defaults to `false`; enabling it can expose proxy listeners beyond localhost.",

660 660 },

661Enable Smart approvals (`prefix_rule` suggestions on escalation requests; stable; on by default).661 {

662 662 key: "features.network_proxy.dangerously_allow_all_unix_sockets",

663Key663 type: "boolean",

664 664 description:

665`features.runtime_metrics`665 "Permit arbitrary Unix socket destinations instead of allowlist-only access. Defaults to `false`; use only in tightly controlled environments.",

666 666 },

667Type / Values667 {

668 668 key: "features.network_proxy.proxy_url",

669`boolean`669 type: "string",

670 670 description:

671Details671 'HTTP listener URL for sandboxed networking. Defaults to `"http://127.0.0.1:3128"`.',

672 672 },

673Show runtime metrics summary in TUI turn separators (experimental).673 {

674 674 key: "features.network_proxy.socks_url",

675Key675 type: "string",

676 676 description:

677`features.search_tool`677 'SOCKS5 listener URL. Defaults to `"http://127.0.0.1:8081"`.',

678 678 },

679Type / Values679 {

680 680 key: "features.web_search",

681`boolean`681 type: "boolean",

682 682 description:

683Details683 "Deprecated legacy toggle; prefer the top-level `web_search` setting.",

684 684 },

685Enable `search_tool_bm25` for Apps tool discovery before invoking app MCP tools (experimental).685 {

686 686 key: "features.web_search_cached",

687Key687 type: "boolean",

688 688 description:

689`features.shell_snapshot`689 'Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`.',

690 690 },

691Type / Values691 {

692 692 key: "features.web_search_request",

693`boolean`693 type: "boolean",

694 694 description:

695Details695 'Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`.',

696 696 },

697Snapshot shell environment to speed up repeated commands (beta).697 {

698 698 key: "features.shell_tool",

699Key699 type: "boolean",

700 700 description:

701`features.shell_tool`701 "Enable the default `shell` tool for running commands (stable; on by default).",

702 702 },

703Type / Values703 {

704 704 key: "features.enable_request_compression",

705`boolean`705 type: "boolean",

706 706 description:

707Details707 "Compress streaming request bodies with zstd when supported (stable; on by default).",

708 708 },

709Enable the default `shell` tool for running commands (stable; on by default).709 {

710 710 key: "features.skill_mcp_dependency_install",

711Key711 type: "boolean",

712 712 description:

713`features.unified_exec`713 "Allow prompting and installing missing MCP dependencies for skills (stable; on by default).",

714 714 },

715Type / Values715 {

716 716 key: "features.fast_mode",

717`boolean`717 type: "boolean",

718 718 description:

719Details719 "Enable model-catalog service tier selection in the TUI, including Fast-tier commands when the active model advertises them (stable; on by default).",

720 720 },

721Use the unified PTY-backed exec tool (beta).721 {

722 722 key: "features.prevent_idle_sleep",

723Key723 type: "boolean",

724 724 description:

725`features.use_linux_sandbox_bwrap`725 "Prevent the machine from sleeping while a turn is actively running (experimental; off by default).",

726 726 },

727Type / Values727 {

728 728 key: "suppress_unstable_features_warning",

729`boolean`729 type: "boolean",

730 730 description:

731Details731 "Suppress the warning that appears when under-development feature flags are enabled.",

732 732 },

733Use the bubblewrap-based Linux sandbox pipeline (experimental; off by default).733 {

734 734 key: "model_providers.<id>",

735Key735 type: "table",

736 736 description:

737`features.web_search`737 "Custom provider definition. Built-in provider IDs (`openai`, `ollama`, and `lmstudio`) are reserved and cannot be overridden.",

738 738 },

739Type / Values739 {

740 740 key: "model_providers.<id>.name",

741`boolean`741 type: "string",

742 742 description: "Display name for a custom model provider.",

743Details743 },

744 744 {

745Deprecated legacy toggle; prefer the top-level `web_search` setting.745 key: "model_providers.<id>.base_url",

746 746 type: "string",

747Key747 description: "API base URL for the model provider.",

748 748 },

749`features.web_search_cached`749 {

750 750 key: "model_providers.<id>.env_key",

751Type / Values751 type: "string",

752 752 description: "Environment variable supplying the provider API key.",

753`boolean`753 },

754 754 {

755Details755 key: "model_providers.<id>.env_key_instructions",

756 756 type: "string",

757Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "cached"`.757 description: "Optional setup guidance for the provider API key.",

758 758 },

759Key759 {

760 760 key: "model_providers.<id>.experimental_bearer_token",

761`features.web_search_request`761 type: "string",

762 762 description:

763Type / Values763 "Direct bearer token for the provider (discouraged; use `env_key`).",

764 764 },

765`boolean`765 {

766 766 key: "model_providers.<id>.requires_openai_auth",

767Details767 type: "boolean",

768 768 description:

769Deprecated legacy toggle. When `web_search` is unset, true maps to `web_search = "live"`.769 "The provider uses OpenAI authentication (defaults to false).",

770 770 },

771Key771 {

772 772 key: "model_providers.<id>.wire_api",

773`feedback.enabled`773 type: "responses",

774 774 description:

775Type / Values775 "Protocol used by the provider. `responses` is the only supported value, and it is the default when omitted.",

776 776 },

777`boolean`777 {

778 778 key: "model_providers.<id>.query_params",

779Details779 type: "map<string,string>",

780 780 description: "Extra query parameters appended to provider requests.",

781Enable feedback submission via `/feedback` across Codex surfaces (default: true).781 },

782 782 {

783Key783 key: "model_providers.<id>.http_headers",

784 784 type: "map<string,string>",

785`file_opener`785 description: "Static HTTP headers added to provider requests.",

786 786 },

787Type / Values787 {

788 788 key: "model_providers.<id>.env_http_headers",

789`vscode | vscode-insiders | windsurf | cursor | none`789 type: "map<string,string>",

790 790 description:

791Details791 "HTTP headers populated from environment variables when present.",

792 792 },

793URI scheme used to open citations from Codex output (default: `vscode`).793 {

794 794 key: "model_providers.<id>.request_max_retries",

795Key795 type: "number",

796 796 description:

797`forced_chatgpt_workspace_id`797 "Retry count for HTTP requests to the provider (default: 4).",

798 798 },

799Type / Values799 {

800 800 key: "model_providers.<id>.stream_max_retries",

801`string (uuid)`801 type: "number",

802 802 description: "Retry count for SSE streaming interruptions (default: 5).",

803Details803 },

804 804 {

805Limit ChatGPT logins to a specific workspace identifier.805 key: "model_providers.<id>.stream_idle_timeout_ms",

806 806 type: "number",

807Key807 description:

808 808 "Idle timeout for SSE streams in milliseconds (default: 300000).",

809`forced_login_method`809 },

810 810 {

811Type / Values811 key: "model_providers.<id>.supports_websockets",

812 812 type: "boolean",

813`chatgpt | api`813 description:

814 814 "Whether that provider supports the Responses API WebSocket transport.",

815Details815 },

816 816 {

817Restrict Codex to a specific authentication method.817 key: "model_providers.<id>.auth",

818 818 type: "table",

819Key819 description:

820 820 "Command-backed bearer token configuration for a custom provider. Do not combine with `env_key`, `experimental_bearer_token`, or `requires_openai_auth`.",

821`hide_agent_reasoning`821 },

822 822 {

823Type / Values823 key: "model_providers.<id>.auth.command",

824 824 type: "string",

825`boolean`825 description:

826 826 "Command to run when Codex needs a bearer token. The command must print the token to stdout.",

827Details827 },

828 828 {

829Suppress reasoning events in both the TUI and `codex exec` output.829 key: "model_providers.<id>.auth.args",

830 830 type: "array<string>",

831Key831 description: "Arguments passed to the token command.",

832 832 },

833`history.max_bytes`833 {

834 834 key: "model_providers.<id>.auth.timeout_ms",