Files
codex/codex-rs/app-server-protocol/schema/json/v2/ConfigReadResponse.json
Charley Cunningham bc24017d64 Add Smart Approvals guardian review across core, app-server, and TUI (#13860)
## Summary
- add `approvals_reviewer = "user" | "guardian_subagent"` as the runtime
control for who reviews approval requests
- route Smart Approvals guardian review through core for command
execution, file changes, managed-network approvals, MCP approvals, and
delegated/subagent approval flows
- expose guardian review in app-server with temporary unstable
`item/autoApprovalReview/{started,completed}` notifications carrying
`targetItemId`, `review`, and `action`
- update the TUI so Smart Approvals can be enabled from `/experimental`,
aligned with the matching `/approvals` mode, and surfaced clearly while
reviews are pending or resolved

## Runtime model
This PR does not introduce a new `approval_policy`.

Instead:
- `approval_policy` still controls when approval is needed
- `approvals_reviewer` controls who reviewable approval requests are
routed to:
  - `user`
  - `guardian_subagent`

`guardian_subagent` is a carefully prompted reviewer subagent that
gathers relevant context and applies a risk-based decision framework
before approving or denying the request.

The `smart_approvals` feature flag is a rollout/UI gate. Core runtime
behavior keys off `approvals_reviewer`.

When Smart Approvals is enabled from the TUI, it also switches the
current `/approvals` settings to the matching Smart Approvals mode so
users immediately see guardian review in the active thread:
- `approval_policy = on-request`
- `approvals_reviewer = guardian_subagent`
- `sandbox_mode = workspace-write`

Users can still change `/approvals` afterward.

Config-load behavior stays intentionally narrow:
- plain `smart_approvals = true` in `config.toml` remains just the
rollout/UI gate and does not auto-set `approvals_reviewer`
- the deprecated `guardian_approval = true` alias migration does
backfill `approvals_reviewer = "guardian_subagent"` in the same scope
when that reviewer is not already configured there, so old configs
preserve their original guardian-enabled behavior

ARC remains a separate safety check. For MCP tool approvals, ARC
escalations now flow into the configured reviewer instead of always
bypassing guardian and forcing manual review.

## Config stability
The runtime reviewer override is stable, but the config-backed
app-server protocol shape is still settling.

- `thread/start`, `thread/resume`, and `turn/start` keep stable
`approvalsReviewer` overrides
- the config-backed `approvals_reviewer` exposure returned via
`config/read` (including profile-level config) is now marked
`[UNSTABLE]` / experimental in the app-server protocol until we are more
confident in that config surface

## App-server surface
This PR intentionally keeps the guardian app-server shape narrow and
temporary.

It adds generic unstable lifecycle notifications:
- `item/autoApprovalReview/started`
- `item/autoApprovalReview/completed`

with payloads of the form:
- `{ threadId, turnId, targetItemId, review, action? }`

`review` is currently:
- `{ status, riskScore?, riskLevel?, rationale? }`
- where `status` is one of `inProgress`, `approved`, `denied`, or
`aborted`

`action` carries the guardian action summary payload from core when
available. This lets clients render temporary standalone pending-review
UI, including parallel reviews, even when the underlying tool item has
not been emitted yet.

These notifications are explicitly documented as `[UNSTABLE]` and
expected to change soon.

This PR does **not** persist guardian review state onto `thread/read`
tool items. The intended follow-up is to attach guardian review state to
the reviewed tool item lifecycle instead, which would improve
consistency with manual approvals and allow thread history / reconnect
flows to replay guardian review state directly.

## TUI behavior
- `/experimental` exposes the rollout gate as `Smart Approvals`
- enabling it in the TUI enables the feature and switches the current
session to the matching Smart Approvals `/approvals` mode
- disabling it in the TUI clears the persisted `approvals_reviewer`
override when appropriate and returns the session to default manual
review when the effective reviewer changes
- `/approvals` still exposes the reviewer choice directly
- the TUI renders:
- pending guardian review state in the live status footer, including
parallel review aggregation
  - resolved approval/denial state in history

## Scope notes
This PR includes the supporting core/runtime work needed to make Smart
Approvals usable end-to-end:
- shell / unified-exec / apply_patch / managed-network / MCP guardian
review
- delegated/subagent approval routing into guardian review
- guardian review risk metadata and action summaries for app-server/TUI
- config/profile/TUI handling for `smart_approvals`, `guardian_approval`
alias migration, and `approvals_reviewer`
- a small internal cleanup of delegated approval forwarding to dedupe
fallback paths and simplify guardian-vs-parent approval waiting (no
intended behavior change)

Out of scope for this PR:
- redesigning the existing manual approval protocol shapes
- persisting guardian review state onto app-server `ThreadItem`s
- delegated MCP elicitation auto-review (the current delegated MCP
guardian shim only covers the legacy `RequestUserInput` path)

---------

Co-authored-by: Codex <noreply@openai.com>
2026-03-13 15:27:00 -07:00

901 lines
21 KiB
JSON

{
"$schema": "http://json-schema.org/draft-07/schema#",
"definitions": {
"AbsolutePathBuf": {
"description": "A path that is guaranteed to be absolute and normalized (though it is not guaranteed to be canonicalized or exist on the filesystem).\n\nIMPORTANT: When deserializing an `AbsolutePathBuf`, a base path must be set using [AbsolutePathBufGuard::new]. If no base path is set, the deserialization will fail unless the path being deserialized is already absolute.",
"type": "string"
},
"AnalyticsConfig": {
"additionalProperties": true,
"properties": {
"enabled": {
"type": [
"boolean",
"null"
]
}
},
"type": "object"
},
"AppConfig": {
"properties": {
"default_tools_approval_mode": {
"anyOf": [
{
"$ref": "#/definitions/AppToolApproval"
},
{
"type": "null"
}
]
},
"default_tools_enabled": {
"type": [
"boolean",
"null"
]
},
"destructive_enabled": {
"type": [
"boolean",
"null"
]
},
"enabled": {
"default": true,
"type": "boolean"
},
"open_world_enabled": {
"type": [
"boolean",
"null"
]
},
"tools": {
"anyOf": [
{
"$ref": "#/definitions/AppToolsConfig"
},
{
"type": "null"
}
]
}
},
"type": "object"
},
"AppToolApproval": {
"enum": [
"auto",
"prompt",
"approve"
],
"type": "string"
},
"AppToolConfig": {
"properties": {
"approval_mode": {
"anyOf": [
{
"$ref": "#/definitions/AppToolApproval"
},
{
"type": "null"
}
]
},
"enabled": {
"type": [
"boolean",
"null"
]
}
},
"type": "object"
},
"AppToolsConfig": {
"type": "object"
},
"ApprovalsReviewer": {
"description": "Configures who approval requests are routed to for review. Examples include sandbox escapes, blocked network access, MCP approval prompts, and ARC escalations. Defaults to `user`. `guardian_subagent` uses a carefully prompted subagent to gather relevant context and apply a risk-based decision framework before approving or denying the request.",
"enum": [
"user",
"guardian_subagent"
],
"type": "string"
},
"AppsConfig": {
"properties": {
"_default": {
"anyOf": [
{
"$ref": "#/definitions/AppsDefaultConfig"
},
{
"type": "null"
}
],
"default": null
}
},
"type": "object"
},
"AppsDefaultConfig": {
"properties": {
"destructive_enabled": {
"default": true,
"type": "boolean"
},
"enabled": {
"default": true,
"type": "boolean"
},
"open_world_enabled": {
"default": true,
"type": "boolean"
}
},
"type": "object"
},
"AskForApproval": {
"oneOf": [
{
"enum": [
"untrusted",
"on-failure",
"on-request",
"never"
],
"type": "string"
},
{
"additionalProperties": false,
"properties": {
"granular": {
"properties": {
"mcp_elicitations": {
"type": "boolean"
},
"request_permissions": {
"default": false,
"type": "boolean"
},
"rules": {
"type": "boolean"
},
"sandbox_approval": {
"type": "boolean"
},
"skill_approval": {
"default": false,
"type": "boolean"
}
},
"required": [
"mcp_elicitations",
"rules",
"sandbox_approval"
],
"type": "object"
}
},
"required": [
"granular"
],
"title": "GranularAskForApproval",
"type": "object"
}
]
},
"Config": {
"additionalProperties": true,
"properties": {
"analytics": {
"anyOf": [
{
"$ref": "#/definitions/AnalyticsConfig"
},
{
"type": "null"
}
]
},
"approval_policy": {
"anyOf": [
{
"$ref": "#/definitions/AskForApproval"
},
{
"type": "null"
}
]
},
"approvals_reviewer": {
"anyOf": [
{
"$ref": "#/definitions/ApprovalsReviewer"
},
{
"type": "null"
}
],
"description": "[UNSTABLE] Optional default for where approval requests are routed for review."
},
"compact_prompt": {
"type": [
"string",
"null"
]
},
"developer_instructions": {
"type": [
"string",
"null"
]
},
"forced_chatgpt_workspace_id": {
"type": [
"string",
"null"
]
},
"forced_login_method": {
"anyOf": [
{
"$ref": "#/definitions/ForcedLoginMethod"
},
{
"type": "null"
}
]
},
"instructions": {
"type": [
"string",
"null"
]
},
"model": {
"type": [
"string",
"null"
]
},
"model_auto_compact_token_limit": {
"format": "int64",
"type": [
"integer",
"null"
]
},
"model_context_window": {
"format": "int64",
"type": [
"integer",
"null"
]
},
"model_provider": {
"type": [
"string",
"null"
]
},
"model_reasoning_effort": {
"anyOf": [
{
"$ref": "#/definitions/ReasoningEffort"
},
{
"type": "null"
}
]
},
"model_reasoning_summary": {
"anyOf": [
{
"$ref": "#/definitions/ReasoningSummary"
},
{
"type": "null"
}
]
},
"model_verbosity": {
"anyOf": [
{
"$ref": "#/definitions/Verbosity"
},
{
"type": "null"
}
]
},
"profile": {
"type": [
"string",
"null"
]
},
"profiles": {
"additionalProperties": {
"$ref": "#/definitions/ProfileV2"
},
"default": {},
"type": "object"
},
"review_model": {
"type": [
"string",
"null"
]
},
"sandbox_mode": {
"anyOf": [
{
"$ref": "#/definitions/SandboxMode"
},
{
"type": "null"
}
]
},
"sandbox_workspace_write": {
"anyOf": [
{
"$ref": "#/definitions/SandboxWorkspaceWrite"
},
{
"type": "null"
}
]
},
"service_tier": {
"anyOf": [
{
"$ref": "#/definitions/ServiceTier"
},
{
"type": "null"
}
]
},
"tools": {
"anyOf": [
{
"$ref": "#/definitions/ToolsV2"
},
{
"type": "null"
}
]
},
"web_search": {
"anyOf": [
{
"$ref": "#/definitions/WebSearchMode"
},
{
"type": "null"
}
]
}
},
"type": "object"
},
"ConfigLayer": {
"properties": {
"config": true,
"disabledReason": {
"type": [
"string",
"null"
]
},
"name": {
"$ref": "#/definitions/ConfigLayerSource"
},
"version": {
"type": "string"
}
},
"required": [
"config",
"name",
"version"
],
"type": "object"
},
"ConfigLayerMetadata": {
"properties": {
"name": {
"$ref": "#/definitions/ConfigLayerSource"
},
"version": {
"type": "string"
}
},
"required": [
"name",
"version"
],
"type": "object"
},
"ConfigLayerSource": {
"oneOf": [
{
"description": "Managed preferences layer delivered by MDM (macOS only).",
"properties": {
"domain": {
"type": "string"
},
"key": {
"type": "string"
},
"type": {
"enum": [
"mdm"
],
"title": "MdmConfigLayerSourceType",
"type": "string"
}
},
"required": [
"domain",
"key",
"type"
],
"title": "MdmConfigLayerSource",
"type": "object"
},
{
"description": "Managed config layer from a file (usually `managed_config.toml`).",
"properties": {
"file": {
"allOf": [
{
"$ref": "#/definitions/AbsolutePathBuf"
}
],
"description": "This is the path to the system config.toml file, though it is not guaranteed to exist."
},
"type": {
"enum": [
"system"
],
"title": "SystemConfigLayerSourceType",
"type": "string"
}
},
"required": [
"file",
"type"
],
"title": "SystemConfigLayerSource",
"type": "object"
},
{
"description": "User config layer from $CODEX_HOME/config.toml. This layer is special in that it is expected to be: - writable by the user - generally outside the workspace directory",
"properties": {
"file": {
"allOf": [
{
"$ref": "#/definitions/AbsolutePathBuf"
}
],
"description": "This is the path to the user's config.toml file, though it is not guaranteed to exist."
},
"type": {
"enum": [
"user"
],
"title": "UserConfigLayerSourceType",
"type": "string"
}
},
"required": [
"file",
"type"
],
"title": "UserConfigLayerSource",
"type": "object"
},
{
"description": "Path to a .codex/ folder within a project. There could be multiple of these between `cwd` and the project/repo root.",
"properties": {
"dotCodexFolder": {
"$ref": "#/definitions/AbsolutePathBuf"
},
"type": {
"enum": [
"project"
],
"title": "ProjectConfigLayerSourceType",
"type": "string"
}
},
"required": [
"dotCodexFolder",
"type"
],
"title": "ProjectConfigLayerSource",
"type": "object"
},
{
"description": "Session-layer overrides supplied via `-c`/`--config`.",
"properties": {
"type": {
"enum": [
"sessionFlags"
],
"title": "SessionFlagsConfigLayerSourceType",
"type": "string"
}
},
"required": [
"type"
],
"title": "SessionFlagsConfigLayerSource",
"type": "object"
},
{
"description": "`managed_config.toml` was designed to be a config that was loaded as the last layer on top of everything else. This scheme did not quite work out as intended, but we keep this variant as a \"best effort\" while we phase out `managed_config.toml` in favor of `requirements.toml`.",
"properties": {
"file": {
"$ref": "#/definitions/AbsolutePathBuf"
},
"type": {
"enum": [
"legacyManagedConfigTomlFromFile"
],
"title": "LegacyManagedConfigTomlFromFileConfigLayerSourceType",
"type": "string"
}
},
"required": [
"file",
"type"
],
"title": "LegacyManagedConfigTomlFromFileConfigLayerSource",
"type": "object"
},
{
"properties": {
"type": {
"enum": [
"legacyManagedConfigTomlFromMdm"
],
"title": "LegacyManagedConfigTomlFromMdmConfigLayerSourceType",
"type": "string"
}
},
"required": [
"type"
],
"title": "LegacyManagedConfigTomlFromMdmConfigLayerSource",
"type": "object"
}
]
},
"ForcedLoginMethod": {
"enum": [
"chatgpt",
"api"
],
"type": "string"
},
"ProfileV2": {
"additionalProperties": true,
"properties": {
"approval_policy": {
"anyOf": [
{
"$ref": "#/definitions/AskForApproval"
},
{
"type": "null"
}
]
},
"approvals_reviewer": {
"anyOf": [
{
"$ref": "#/definitions/ApprovalsReviewer"
},
{
"type": "null"
}
],
"description": "[UNSTABLE] Optional profile-level override for where approval requests are routed for review. If omitted, the enclosing config default is used."
},
"chatgpt_base_url": {
"type": [
"string",
"null"
]
},
"model": {
"type": [
"string",
"null"
]
},
"model_provider": {
"type": [
"string",
"null"
]
},
"model_reasoning_effort": {
"anyOf": [
{
"$ref": "#/definitions/ReasoningEffort"
},
{
"type": "null"
}
]
},
"model_reasoning_summary": {
"anyOf": [
{
"$ref": "#/definitions/ReasoningSummary"
},
{
"type": "null"
}
]
},
"model_verbosity": {
"anyOf": [
{
"$ref": "#/definitions/Verbosity"
},
{
"type": "null"
}
]
},
"service_tier": {
"anyOf": [
{
"$ref": "#/definitions/ServiceTier"
},
{
"type": "null"
}
]
},
"tools": {
"anyOf": [
{
"$ref": "#/definitions/ToolsV2"
},
{
"type": "null"
}
]
},
"web_search": {
"anyOf": [
{
"$ref": "#/definitions/WebSearchMode"
},
{
"type": "null"
}
]
}
},
"type": "object"
},
"ReasoningEffort": {
"description": "See https://platform.openai.com/docs/guides/reasoning?api-mode=responses#get-started-with-reasoning",
"enum": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
],
"type": "string"
},
"ReasoningSummary": {
"description": "A summary of the reasoning performed by the model. This can be useful for debugging and understanding the model's reasoning process. See https://platform.openai.com/docs/guides/reasoning?api-mode=responses#reasoning-summaries",
"oneOf": [
{
"enum": [
"auto",
"concise",
"detailed"
],
"type": "string"
},
{
"description": "Option to disable reasoning summaries.",
"enum": [
"none"
],
"type": "string"
}
]
},
"SandboxMode": {
"enum": [
"read-only",
"workspace-write",
"danger-full-access"
],
"type": "string"
},
"SandboxWorkspaceWrite": {
"properties": {
"exclude_slash_tmp": {
"default": false,
"type": "boolean"
},
"exclude_tmpdir_env_var": {
"default": false,
"type": "boolean"
},
"network_access": {
"default": false,
"type": "boolean"
},
"writable_roots": {
"default": [],
"items": {
"type": "string"
},
"type": "array"
}
},
"type": "object"
},
"ServiceTier": {
"enum": [
"fast",
"flex"
],
"type": "string"
},
"ToolsV2": {
"properties": {
"view_image": {
"type": [
"boolean",
"null"
]
},
"web_search": {
"anyOf": [
{
"$ref": "#/definitions/WebSearchToolConfig"
},
{
"type": "null"
}
]
}
},
"type": "object"
},
"Verbosity": {
"description": "Controls output length/detail on GPT-5 models via the Responses API. Serialized with lowercase values to match the OpenAI API.",
"enum": [
"low",
"medium",
"high"
],
"type": "string"
},
"WebSearchContextSize": {
"enum": [
"low",
"medium",
"high"
],
"type": "string"
},
"WebSearchLocation": {
"additionalProperties": false,
"properties": {
"city": {
"type": [
"string",
"null"
]
},
"country": {
"type": [
"string",
"null"
]
},
"region": {
"type": [
"string",
"null"
]
},
"timezone": {
"type": [
"string",
"null"
]
}
},
"type": "object"
},
"WebSearchMode": {
"enum": [
"disabled",
"cached",
"live"
],
"type": "string"
},
"WebSearchToolConfig": {
"additionalProperties": false,
"properties": {
"allowed_domains": {
"items": {
"type": "string"
},
"type": [
"array",
"null"
]
},
"context_size": {
"anyOf": [
{
"$ref": "#/definitions/WebSearchContextSize"
},
{
"type": "null"
}
]
},
"location": {
"anyOf": [
{
"$ref": "#/definitions/WebSearchLocation"
},
{
"type": "null"
}
]
}
},
"type": "object"
}
},
"properties": {
"config": {
"$ref": "#/definitions/Config"
},
"layers": {
"items": {
"$ref": "#/definitions/ConfigLayer"
},
"type": [
"array",
"null"
]
},
"origins": {
"additionalProperties": {
"$ref": "#/definitions/ConfigLayerMetadata"
},
"type": "object"
}
},
"required": [
"config",
"origins"
],
"title": "ConfigReadResponse",
"type": "object"
}