Files
codex/codex-rs/app-server-protocol/schema/json/v2/ThreadResumeParams.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

1108 lines
26 KiB
JSON

{
"$schema": "http://json-schema.org/draft-07/schema#",
"definitions": {
"ApprovalsReviewer": {
"description": "Configures who approval requests are routed to for review. Examples include sandbox escapes, blocked network access, MCP approval prompts, and ARC escalations. Defaults to `user`. `guardian_subagent` uses a carefully prompted subagent to gather relevant context and apply a risk-based decision framework before approving or denying the request.",
"enum": [
"user",
"guardian_subagent"
],
"type": "string"
},
"AskForApproval": {
"oneOf": [
{
"enum": [
"untrusted",
"on-failure",
"on-request",
"never"
],
"type": "string"
},
{
"additionalProperties": false,
"properties": {
"granular": {
"properties": {
"mcp_elicitations": {
"type": "boolean"
},
"request_permissions": {
"default": false,
"type": "boolean"
},
"rules": {
"type": "boolean"
},
"sandbox_approval": {
"type": "boolean"
},
"skill_approval": {
"default": false,
"type": "boolean"
}
},
"required": [
"mcp_elicitations",
"rules",
"sandbox_approval"
],
"type": "object"
}
},
"required": [
"granular"
],
"title": "GranularAskForApproval",
"type": "object"
}
]
},
"ContentItem": {
"oneOf": [
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"input_text"
],
"title": "InputTextContentItemType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "InputTextContentItem",
"type": "object"
},
{
"properties": {
"image_url": {
"type": "string"
},
"type": {
"enum": [
"input_image"
],
"title": "InputImageContentItemType",
"type": "string"
}
},
"required": [
"image_url",
"type"
],
"title": "InputImageContentItem",
"type": "object"
},
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"output_text"
],
"title": "OutputTextContentItemType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "OutputTextContentItem",
"type": "object"
}
]
},
"FunctionCallOutputBody": {
"anyOf": [
{
"type": "string"
},
{
"items": {
"$ref": "#/definitions/FunctionCallOutputContentItem"
},
"type": "array"
}
]
},
"FunctionCallOutputContentItem": {
"description": "Responses API compatible content items that can be returned by a tool call. This is a subset of ContentItem with the types we support as function call outputs.",
"oneOf": [
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"input_text"
],
"title": "InputTextFunctionCallOutputContentItemType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "InputTextFunctionCallOutputContentItem",
"type": "object"
},
{
"properties": {
"detail": {
"anyOf": [
{
"$ref": "#/definitions/ImageDetail"
},
{
"type": "null"
}
]
},
"image_url": {
"type": "string"
},
"type": {
"enum": [
"input_image"
],
"title": "InputImageFunctionCallOutputContentItemType",
"type": "string"
}
},
"required": [
"image_url",
"type"
],
"title": "InputImageFunctionCallOutputContentItem",
"type": "object"
}
]
},
"FunctionCallOutputPayload": {
"description": "The payload we send back to OpenAI when reporting a tool call result.\n\n`body` serializes directly as the wire value for `function_call_output.output`. `success` remains internal metadata for downstream handling.",
"properties": {
"body": {
"$ref": "#/definitions/FunctionCallOutputBody"
},
"success": {
"type": [
"boolean",
"null"
]
}
},
"required": [
"body"
],
"type": "object"
},
"GhostCommit": {
"description": "Details of a ghost commit created from a repository state.",
"properties": {
"id": {
"type": "string"
},
"parent": {
"type": [
"string",
"null"
]
},
"preexisting_untracked_dirs": {
"items": {
"type": "string"
},
"type": "array"
},
"preexisting_untracked_files": {
"items": {
"type": "string"
},
"type": "array"
}
},
"required": [
"id",
"preexisting_untracked_dirs",
"preexisting_untracked_files"
],
"type": "object"
},
"ImageDetail": {
"enum": [
"auto",
"low",
"high",
"original"
],
"type": "string"
},
"LocalShellAction": {
"oneOf": [
{
"properties": {
"command": {
"items": {
"type": "string"
},
"type": "array"
},
"env": {
"additionalProperties": {
"type": "string"
},
"type": [
"object",
"null"
]
},
"timeout_ms": {
"format": "uint64",
"minimum": 0.0,
"type": [
"integer",
"null"
]
},
"type": {
"enum": [
"exec"
],
"title": "ExecLocalShellActionType",
"type": "string"
},
"user": {
"type": [
"string",
"null"
]
},
"working_directory": {
"type": [
"string",
"null"
]
}
},
"required": [
"command",
"type"
],
"title": "ExecLocalShellAction",
"type": "object"
}
]
},
"LocalShellStatus": {
"enum": [
"completed",
"in_progress",
"incomplete"
],
"type": "string"
},
"MessagePhase": {
"description": "Classifies an assistant message as interim commentary or final answer text.\n\nProviders do not emit this consistently, so callers must treat `None` as \"phase unknown\" and keep compatibility behavior for legacy models.",
"oneOf": [
{
"description": "Mid-turn assistant text (for example preamble/progress narration).\n\nAdditional tool calls or assistant output may follow before turn completion.",
"enum": [
"commentary"
],
"type": "string"
},
{
"description": "The assistant's terminal answer text for the current turn.",
"enum": [
"final_answer"
],
"type": "string"
}
]
},
"Personality": {
"enum": [
"none",
"friendly",
"pragmatic"
],
"type": "string"
},
"ReasoningItemContent": {
"oneOf": [
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"reasoning_text"
],
"title": "ReasoningTextReasoningItemContentType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "ReasoningTextReasoningItemContent",
"type": "object"
},
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"text"
],
"title": "TextReasoningItemContentType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "TextReasoningItemContent",
"type": "object"
}
]
},
"ReasoningItemReasoningSummary": {
"oneOf": [
{
"properties": {
"text": {
"type": "string"
},
"type": {
"enum": [
"summary_text"
],
"title": "SummaryTextReasoningItemReasoningSummaryType",
"type": "string"
}
},
"required": [
"text",
"type"
],
"title": "SummaryTextReasoningItemReasoningSummary",
"type": "object"
}
]
},
"ResponseItem": {
"oneOf": [
{
"properties": {
"content": {
"items": {
"$ref": "#/definitions/ContentItem"
},
"type": "array"
},
"end_turn": {
"type": [
"boolean",
"null"
]
},
"id": {
"type": [
"string",
"null"
],
"writeOnly": true
},
"phase": {
"anyOf": [
{
"$ref": "#/definitions/MessagePhase"
},
{
"type": "null"
}
]
},
"role": {
"type": "string"
},
"type": {
"enum": [
"message"
],
"title": "MessageResponseItemType",
"type": "string"
}
},
"required": [
"content",
"role",
"type"
],
"title": "MessageResponseItem",
"type": "object"
},
{
"properties": {
"content": {
"default": null,
"items": {
"$ref": "#/definitions/ReasoningItemContent"
},
"type": [
"array",
"null"
]
},
"encrypted_content": {
"type": [
"string",
"null"
]
},
"id": {
"type": "string",
"writeOnly": true
},
"summary": {
"items": {
"$ref": "#/definitions/ReasoningItemReasoningSummary"
},
"type": "array"
},
"type": {
"enum": [
"reasoning"
],
"title": "ReasoningResponseItemType",
"type": "string"
}
},
"required": [
"id",
"summary",
"type"
],
"title": "ReasoningResponseItem",
"type": "object"
},
{
"properties": {
"action": {
"$ref": "#/definitions/LocalShellAction"
},
"call_id": {
"description": "Set when using the Responses API.",
"type": [
"string",
"null"
]
},
"id": {
"description": "Legacy id field retained for compatibility with older payloads.",
"type": [
"string",
"null"
],
"writeOnly": true
},
"status": {
"$ref": "#/definitions/LocalShellStatus"
},
"type": {
"enum": [
"local_shell_call"
],
"title": "LocalShellCallResponseItemType",
"type": "string"
}
},
"required": [
"action",
"status",
"type"
],
"title": "LocalShellCallResponseItem",
"type": "object"
},
{
"properties": {
"arguments": {
"type": "string"
},
"call_id": {
"type": "string"
},
"id": {
"type": [
"string",
"null"
],
"writeOnly": true
},
"name": {
"type": "string"
},
"namespace": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"function_call"
],
"title": "FunctionCallResponseItemType",
"type": "string"
}
},
"required": [
"arguments",
"call_id",
"name",
"type"
],
"title": "FunctionCallResponseItem",
"type": "object"
},
{
"properties": {
"arguments": true,
"call_id": {
"type": [
"string",
"null"
]
},
"execution": {
"type": "string"
},
"id": {
"type": [
"string",
"null"
],
"writeOnly": true
},
"status": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"tool_search_call"
],
"title": "ToolSearchCallResponseItemType",
"type": "string"
}
},
"required": [
"arguments",
"execution",
"type"
],
"title": "ToolSearchCallResponseItem",
"type": "object"
},
{
"properties": {
"call_id": {
"type": "string"
},
"output": {
"$ref": "#/definitions/FunctionCallOutputPayload"
},
"type": {
"enum": [
"function_call_output"
],
"title": "FunctionCallOutputResponseItemType",
"type": "string"
}
},
"required": [
"call_id",
"output",
"type"
],
"title": "FunctionCallOutputResponseItem",
"type": "object"
},
{
"properties": {
"call_id": {
"type": "string"
},
"id": {
"type": [
"string",
"null"
],
"writeOnly": true
},
"input": {
"type": "string"
},
"name": {
"type": "string"
},
"status": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"custom_tool_call"
],
"title": "CustomToolCallResponseItemType",
"type": "string"
}
},
"required": [
"call_id",
"input",
"name",
"type"
],
"title": "CustomToolCallResponseItem",
"type": "object"
},
{
"properties": {
"call_id": {
"type": "string"
},
"output": {
"$ref": "#/definitions/FunctionCallOutputPayload"
},
"type": {
"enum": [
"custom_tool_call_output"
],
"title": "CustomToolCallOutputResponseItemType",
"type": "string"
}
},
"required": [
"call_id",
"output",
"type"
],
"title": "CustomToolCallOutputResponseItem",
"type": "object"
},
{
"properties": {
"call_id": {
"type": [
"string",
"null"
]
},
"execution": {
"type": "string"
},
"status": {
"type": "string"
},
"tools": {
"items": true,
"type": "array"
},
"type": {
"enum": [
"tool_search_output"
],
"title": "ToolSearchOutputResponseItemType",
"type": "string"
}
},
"required": [
"execution",
"status",
"tools",
"type"
],
"title": "ToolSearchOutputResponseItem",
"type": "object"
},
{
"properties": {
"action": {
"anyOf": [
{
"$ref": "#/definitions/ResponsesApiWebSearchAction"
},
{
"type": "null"
}
]
},
"id": {
"type": [
"string",
"null"
],
"writeOnly": true
},
"status": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"web_search_call"
],
"title": "WebSearchCallResponseItemType",
"type": "string"
}
},
"required": [
"type"
],
"title": "WebSearchCallResponseItem",
"type": "object"
},
{
"properties": {
"id": {
"type": "string"
},
"result": {
"type": "string"
},
"revised_prompt": {
"type": [
"string",
"null"
]
},
"status": {
"type": "string"
},
"type": {
"enum": [
"image_generation_call"
],
"title": "ImageGenerationCallResponseItemType",
"type": "string"
}
},
"required": [
"id",
"result",
"status",
"type"
],
"title": "ImageGenerationCallResponseItem",
"type": "object"
},
{
"properties": {
"ghost_commit": {
"$ref": "#/definitions/GhostCommit"
},
"type": {
"enum": [
"ghost_snapshot"
],
"title": "GhostSnapshotResponseItemType",
"type": "string"
}
},
"required": [
"ghost_commit",
"type"
],
"title": "GhostSnapshotResponseItem",
"type": "object"
},
{
"properties": {
"encrypted_content": {
"type": "string"
},
"type": {
"enum": [
"compaction"
],
"title": "CompactionResponseItemType",
"type": "string"
}
},
"required": [
"encrypted_content",
"type"
],
"title": "CompactionResponseItem",
"type": "object"
},
{
"properties": {
"type": {
"enum": [
"other"
],
"title": "OtherResponseItemType",
"type": "string"
}
},
"required": [
"type"
],
"title": "OtherResponseItem",
"type": "object"
}
]
},
"ResponsesApiWebSearchAction": {
"oneOf": [
{
"properties": {
"queries": {
"items": {
"type": "string"
},
"type": [
"array",
"null"
]
},
"query": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"search"
],
"title": "SearchResponsesApiWebSearchActionType",
"type": "string"
}
},
"required": [
"type"
],
"title": "SearchResponsesApiWebSearchAction",
"type": "object"
},
{
"properties": {
"type": {
"enum": [
"open_page"
],
"title": "OpenPageResponsesApiWebSearchActionType",
"type": "string"
},
"url": {
"type": [
"string",
"null"
]
}
},
"required": [
"type"
],
"title": "OpenPageResponsesApiWebSearchAction",
"type": "object"
},
{
"properties": {
"pattern": {
"type": [
"string",
"null"
]
},
"type": {
"enum": [
"find_in_page"
],
"title": "FindInPageResponsesApiWebSearchActionType",
"type": "string"
},
"url": {
"type": [
"string",
"null"
]
}
},
"required": [
"type"
],
"title": "FindInPageResponsesApiWebSearchAction",
"type": "object"
},
{
"properties": {
"type": {
"enum": [
"other"
],
"title": "OtherResponsesApiWebSearchActionType",
"type": "string"
}
},
"required": [
"type"
],
"title": "OtherResponsesApiWebSearchAction",
"type": "object"
}
]
},
"SandboxMode": {
"enum": [
"read-only",
"workspace-write",
"danger-full-access"
],
"type": "string"
},
"ServiceTier": {
"enum": [
"fast",
"flex"
],
"type": "string"
}
},
"description": "There are three ways to resume a thread: 1. By thread_id: load the thread from disk by thread_id and resume it. 2. By history: instantiate the thread from memory and resume it. 3. By path: load the thread from disk by path and resume it.\n\nThe precedence is: history > path > thread_id. If using history or path, the thread_id param will be ignored.\n\nPrefer using thread_id whenever possible.",
"properties": {
"approvalPolicy": {
"anyOf": [
{
"$ref": "#/definitions/AskForApproval"
},
{
"type": "null"
}
]
},
"approvalsReviewer": {
"anyOf": [
{
"$ref": "#/definitions/ApprovalsReviewer"
},
{
"type": "null"
}
],
"description": "Override where approval requests are routed for review on this thread and subsequent turns."
},
"baseInstructions": {
"type": [
"string",
"null"
]
},
"config": {
"additionalProperties": true,
"type": [
"object",
"null"
]
},
"cwd": {
"type": [
"string",
"null"
]
},
"developerInstructions": {
"type": [
"string",
"null"
]
},
"model": {
"description": "Configuration overrides for the resumed thread, if any.",
"type": [
"string",
"null"
]
},
"modelProvider": {
"type": [
"string",
"null"
]
},
"personality": {
"anyOf": [
{
"$ref": "#/definitions/Personality"
},
{
"type": "null"
}
]
},
"sandbox": {
"anyOf": [
{
"$ref": "#/definitions/SandboxMode"
},
{
"type": "null"
}
]
},
"serviceTier": {
"anyOf": [
{
"anyOf": [
{
"$ref": "#/definitions/ServiceTier"
},
{
"type": "null"
}
]
},
{
"type": "null"
}
]
},
"threadId": {
"type": "string"
}
},
"required": [
"threadId"
],
"title": "ThreadResumeParams",
"type": "object"
}