enterprise/managed-configuration.md +312 −0 added
1# Managed configuration
2
3Enterprise admins can control local Codex behavior in two ways:
4
5- **Requirements**: admin-enforced constraints that users can't override.
6- **Managed defaults**: starting values applied when Codex launches. Users can still change settings during a session; Codex reapplies managed defaults the next time it starts.
7
8## Admin-enforced requirements (requirements.toml)
9
10Requirements constrain security-sensitive settings (approval policy, approvals reviewer, automatic review policy, sandbox mode, web search mode, managed hooks, and optionally which MCP servers users can enable). When resolving configuration (for example from `config.toml`, profiles, or CLI config overrides), if a value conflicts with an enforced rule, Codex falls back to a compatible value and notifies the user. If you configure an `mcp_servers` allowlist, Codex enables an MCP server only when both its name and identity match an approved entry; otherwise, Codex disables it.
11
12Requirements can also constrain [feature flags](https://developers.openai.com/codex/config-basic/#feature-flags) via the `[features]` table in `requirements.toml`. Note that features aren't always security-sensitive, but enterprises can pin values if desired. Omitted keys remain unconstrained.
13
14For the exact key list, see the [`requirements.toml` section in Configuration Reference](https://developers.openai.com/codex/config-reference#requirementstoml).
15
16### Locations and precedence
17
18Codex applies requirements layers in this order (earlier wins per field):
19
201. Cloud-managed requirements (ChatGPT Business or Enterprise)
212. macOS managed preferences (MDM) via `com.openai.codex:requirements_toml_base64`
223. System `requirements.toml` (`/etc/codex/requirements.toml` on Unix systems, including Linux/macOS, or `%ProgramData%\OpenAI\Codex\requirements.toml` on Windows)
23
24Across layers, Codex merges requirements per field: if an earlier layer sets a field (including an empty list), later layers don't override that field, but lower layers can still fill fields that remain unset.
25
26For backwards compatibility, Codex also interprets legacy `managed_config.toml` fields `approval_policy` and `sandbox_mode` as requirements (allowing only that single value).
27
28### Cloud-managed requirements
29
30When you sign in with ChatGPT on a Business or Enterprise plan, Codex can also fetch admin-enforced requirements from the Codex service. This is another source of `requirements.toml`-compatible requirements. This applies across Codex surfaces, including the CLI, App, and IDE Extension.
31
32#### Configure cloud-managed requirements
33
34Go to the [Codex managed-config page](https://chatgpt.com/codex/settings/managed-configs).
35
36Create a new managed requirements file using the same format and keys as `requirements.toml`.
37
38```toml
39enforce_residency = "us"
40allowed_approval_policies = ["on-request"]
41allowed_sandbox_modes = ["read-only", "workspace-write"]
42
43[rules]
44prefix_rules = [
45 { pattern = [{ any_of = ["bash", "sh", "zsh"] }], decision = "prompt", justification = "Require explicit approval for shell entrypoints" },
46]
47```
48
49Save the configuration. Once saved, the updated managed requirements apply immediately for matching users.
50For more examples, see [Example requirements.toml](#example-requirementstoml).
51
52#### Assign requirements to groups
53
54Admins can configure different managed requirements for different user groups, and also set a default fallback requirements policy.
55
56If a user matches more than one group-specific rule, the first matching rule applies. Codex doesn't fill unset fields from later matching group rules.
57
58For example, if the first matching group rule sets only `allowed_sandbox_modes = ["read-only"]` and a later matching group rule sets `allowed_approval_policies = ["on-request"]`, Codex applies only the first matching group rule and doesn't fill `allowed_approval_policies` from the later rule.
59
60#### How Codex applies cloud-managed requirements locally
61
62When a user starts Codex and signs in with ChatGPT on a Business or Enterprise plan, Codex applies managed requirements on a best-effort basis. Codex first checks for a valid, unexpired local managed requirements cache entry and uses it if available. If the cache is missing, expired, corrupted, or doesn't match the current auth identity, Codex attempts to fetch managed requirements from the service (with retries) and writes a new signed cache entry on success. If no valid cached entry is available and the fetch fails or times out, Codex continues without the managed requirements layer.
63
64After cache resolution, Codex enforces managed requirements as part of the normal requirements layering described above.
65
66### Example requirements.toml
67
68This example blocks `--ask-for-approval never` and `--sandbox danger-full-access` (including `--yolo`):
69
70```toml
71allowed_approval_policies = ["untrusted", "on-request"]
72allowed_sandbox_modes = ["read-only", "workspace-write"]
73```
74
75### Override sandbox requirements by host
76
77Use `[[remote_sandbox_config]]` when one managed policy should apply different
78sandbox requirements on different hosts. For example, you can keep a stricter
79default for laptops while allowing workspace writes on matching devboxes or CI
80runners. Host-specific entries currently override `allowed_sandbox_modes` only:
81
82```toml
83allowed_sandbox_modes = ["read-only"]
84
85[[remote_sandbox_config]]
86hostname_patterns = ["*.devbox.example.com", "runner-??.ci.example.com"]
87allowed_sandbox_modes = ["read-only", "workspace-write"]
88```
89
90Codex compares each `hostname_patterns` entry against the best-effort resolved
91host name. It prefers the fully qualified domain name when available and falls
92back to the local host name. Matching is case-insensitive; `*` matches any
93sequence of characters, and `?` matches one character.
94
95The first matching `[[remote_sandbox_config]]` entry wins within the same
96requirements source. If no entry matches, Codex keeps the top-level
97`allowed_sandbox_modes`. Hostname matching is for policy selection only; don't
98treat it as authenticated device proof.
99
100You can also constrain web search mode:
101
102```toml
103allowed_web_search_modes = ["cached"] # "disabled" remains implicitly allowed
104```
105
106`allowed_web_search_modes = []` allows only `"disabled"`.
107For example, `allowed_web_search_modes = ["cached"]` prevents live web search even in `danger-full-access` sessions.
108
109### Pin feature flags
110
111You can also pin [feature flags](https://developers.openai.com/codex/config-basic/#feature-flags) for users
112receiving a managed `requirements.toml`:
113
114```toml
115[features]
116personality = true
117unified_exec = false
118
119# Disable specific Codex feature surfaces when needed.
120browser_use = false
121in_app_browser = false
122computer_use = false
123```
124
125Use the canonical feature keys from `config.toml`'s `[features]` table. Codex normalizes the resulting feature set to meet these pins and rejects conflicting writes to `config.toml` or profile-scoped feature settings.
126
127- `in_app_browser = false` disables the in-app browser pane.
128- `browser_use = false` disables Browser Use and Browser Agent availability.
129- `computer_use = false` disables Computer Use availability and related
130 install or enablement flows.
131
132If omitted, these features are allowed by policy, subject to normal client,
133platform, and rollout availability.
134
135### Configure automatic review policy
136
137Use `allowed_approvals_reviewers` to require or allow automatic review. Set it
138to `["auto_review"]` to require automatic review, or include `"user"` when users
139can choose manual approval.
140
141Set `guardian_policy_config` to replace the tenant-specific section of the
142automatic review policy. Codex still uses the built-in reviewer template and
143output contract. Managed `guardian_policy_config` takes precedence over local
144`[auto_review].policy`.
145
146```toml
147allowed_approval_policies = ["on-request"]
148allowed_approvals_reviewers = ["auto_review"]
149
150guardian_policy_config = """
151## Environment Profile
152- Trusted internal destinations include github.com/my-org, artifacts.example.com,
153 and internal CI systems.
154
155## Tenant Risk Taxonomy and Allow/Deny Rules
156- Treat uploads to unapproved third-party file-sharing services as high risk.
157- Deny actions that expose credentials or private source code to untrusted
158 destinations.
159"""
160```
161
162### Enforce deny-read requirements
163
164Admins can deny reads for exact paths or glob patterns with
165`[permissions.filesystem]`. Users can't weaken these requirements with local
166configuration.
167
168```toml
169[permissions.filesystem]
170deny_read = [
171 "/Users/alice/.ssh",
172 "./private/**/*.txt",
173]
174```
175
176When deny-read requirements are present, Codex constrains local sandbox mode to
177`read-only` or `workspace-write` so Codex can enforce them. On native
178Windows, managed `deny_read` applies to direct file tools; shell subprocess
179reads don't use this sandbox rule.
180
181### Enforce managed hooks from requirements
182
183Admins can also define managed lifecycle hooks directly in `requirements.toml`.
184Use `[hooks]` for the hook configuration itself, and point `managed_dir` at the
185directory where your MDM or endpoint-management tooling installs the referenced
186scripts.
187
188```toml
189[features]
190codex_hooks = true
191
192[hooks]
193managed_dir = "/enterprise/hooks"
194windows_managed_dir = 'C:\enterprise\hooks'
195
196[[hooks.PreToolUse]]
197matcher = "^Bash$"
198
199[[hooks.PreToolUse.hooks]]
200type = "command"
201command = "python3 /enterprise/hooks/pre_tool_use_policy.py"
202timeout = 30
203statusMessage = "Checking managed Bash command"
204```
205
206Notes:
207
208- Codex enforces the hook configuration from `requirements.toml`, but it does
209 not distribute the scripts in `managed_dir`.
210- Deliver those scripts separately with your MDM or device-management solution.
211- Managed hook commands should reference absolute script paths under the
212 configured managed directory.
213
214### Enforce command rules from requirements
215
216Admins can also enforce restrictive command rules from `requirements.toml`
217using a `[rules]` table. These rules merge with regular `.rules` files, and the
218most restrictive decision still wins.
219
220Unlike `.rules`, requirements rules must specify `decision`, and that decision
221must be `"prompt"` or `"forbidden"` (not `"allow"`).
222
223```toml
224[rules]
225prefix_rules = [
226 { pattern = [{ token = "rm" }], decision = "forbidden", justification = "Use git clean -fd instead." },
227 { pattern = [{ token = "git" }, { any_of = ["push", "commit"] }], decision = "prompt", justification = "Require review before mutating history." },
228]
229```
230
231To restrict which MCP servers Codex can enable, add an `mcp_servers` approved list. For stdio servers, match on `command`; for streamable HTTP servers, match on `url`:
232
233```toml
234[mcp_servers.docs]
235identity = { command = "codex-mcp" }
236
237[mcp_servers.remote]
238identity = { url = "https://example.com/mcp" }
239```
240
241If `mcp_servers` is present but empty, Codex disables all MCP servers.
242
243## Managed defaults (`managed_config.toml`)
244
245Managed defaults merge on top of a user's local `config.toml` and take precedence over any CLI `--config` overrides, setting the starting values when Codex launches. Users can still change those settings during a session; Codex reapplies managed defaults the next time it starts.
246
247Make sure your managed defaults meet your requirements; Codex rejects disallowed values.
248
249### Precedence and layering
250
251Codex assembles the effective configuration in this order (top overrides bottom):
252
253- Managed preferences (macOS MDM; highest precedence)
254- `managed_config.toml` (system/managed file)
255- `config.toml` (user's base configuration)
256
257CLI `--config key=value` overrides apply to the base, but managed layers override them. This means each run starts from the managed defaults even if you provide local flags.
258
259Cloud-managed requirements affect the requirements layer (not managed defaults). See the Admin-enforced requirements section above for precedence.
260
261### Locations
262
263- Linux/macOS (Unix): `/etc/codex/managed_config.toml`
264- Windows/non-Unix: `~/.codex/managed_config.toml`
265
266If the file is missing, Codex skips the managed layer.
267
268### macOS managed preferences (MDM)
269
270On macOS, admins can push a device profile that provides base64-encoded TOML payloads at:
271
272- Preference domain: `com.openai.codex`
273- Keys:
274 - `config_toml_base64` (managed defaults)
275 - `requirements_toml_base64` (requirements)
276
277Codex parses these "managed preferences" payloads as TOML. For managed defaults (`config_toml_base64`), managed preferences have the highest precedence. For requirements (`requirements_toml_base64`), precedence follows the cloud-managed requirements order described above. The same requirements-side `[features]` table works in `requirements_toml_base64`; use canonical feature keys there as well.
278
279### MDM setup workflow
280
281Codex honors standard macOS MDM payloads, so you can distribute settings with tooling like `Jamf Pro`, `Fleet`, or `Kandji`. A lightweight deployment looks like:
282
2831. Build the managed payload TOML and encode it with `base64` (no wrapping).
2842. Drop the string into your MDM profile under the `com.openai.codex` domain at `config_toml_base64` (managed defaults) or `requirements_toml_base64` (requirements).
2853. Push the profile, then ask users to restart Codex and confirm the startup config summary reflects the managed values.
2864. When revoking or changing policy, update the managed payload; the CLI reads the refreshed preference the next time it launches.
287
288Avoid embedding secrets or high-churn dynamic values in the payload. Treat the managed TOML like any other MDM setting under change control.
289
290### Example managed_config.toml
291
292```toml
293# Set conservative defaults
294approval_policy = "on-request"
295sandbox_mode = "workspace-write"
296
297[sandbox_workspace_write]
298network_access = false # keep network disabled unless explicitly allowed
299
300[otel]
301environment = "prod"
302exporter = "otlp-http" # point at your collector
303log_user_prompt = false # keep prompts redacted
304# exporter details live under exporter tables; see Monitoring and telemetry above
305```
306
307### Recommended guardrails
308
309- Prefer `workspace-write` with approvals for most users; reserve full access for controlled containers.
310- Keep `network_access = false` unless your security review allows a collector or domains required by your workflows.
311- Use managed configuration to pin OTel settings (exporter, environment), but keep `log_user_prompt = false` unless your policy explicitly allows storing prompt contents.
312- Periodically audit diffs between local `config.toml` and managed policy to catch drift; managed layers should win over local flags and files.