KaiCalls

Rollback Agent Configuration

rollback_config

DestructiveIdempotent

Request a rollback to a prior assistant_config_versions snapshot. This changes live greeting, voice, prompt and model settings, so authenticated owner review of the exact stored request is required. Caller-supplied authority fields never authorize execution. Discover a target with list_config_versions or get_change_history; relay the pending request's review URL. The agent.config.rollback executor preserves version recording and secret exclusion after trusted approval.

Input Schema

TableJSON Schema

Name	Required	Description
`actor`	No	Who is asking: { type: agent\|human\|system\|integration, id, display_name }.
`reason`	No	Optional human-readable reason recorded in the audit trail.
`dry_run`	No	Validate tenant, scope, and policy without touching the live config.
`agent_id`	Yes	Agent ID to roll back.
`authority`	No	Attribution only. Caller-supplied modes cannot authorize a rollback; authenticated owner review of the exact stored request is required.
`source_ref`	No	External source reference, such as a ticket or automation run ID.
`version_id`	No	Target assistant_config_versions row id to restore. One of version_number/version_id is required.
`business_id`	No	Business ID (optional only when the token can access exactly one business).
`version_number`	No	Target assistant_config_versions.version_number to restore. One of version_number/version_id is required.
`idempotency_key`	No	Stable key for the rollback request; repeating it returns the original outcome. Defaults to a key derived from agent + version — pass a fresh key to repeat a rollback that already executed.
`queue_for_approval`	No	Create a durable owner-review request (default true). Setting false returns needs_approval, but an authority-envelope retry still cannot execute.

Output Schema

TableJSON Schema

Name	Required	Description
`code`	No
`error`	No	Present when success is false
`result`	No	On executed: { agent_id, restored_from_version, restored_fields, skipped_fields, new_version_number, warning? }.
`status`	No
`message`	No
`success`	Yes	Whether the tool completed successfully
`summary`	No
`agent_id`	No
`approval`	No
`request_id`	No
`risk_level`	No
`business_id`	No

Schema Changelog

Changes observed during successful MCP inspections.

Changed2 schema fields changedSep 10, 2026

changedInput schema / properties / authority / description

Previous value: -"How the rollback was authorized: { mode, confirmed_by, confirmed_at, confirmation_ref }. human_confirmed requires all three confirmation fields; anything weaker returns pending_approval."New value: +"Attribution only. Caller-supplied modes cannot authorize a rollback; authenticated owner review of the exact stored request is required."

changedInput schema / properties / queue_for_approval / description

Previous value: -"When authority is insufficient, create a durable dashboard approval (default true). Set false to get needs_approval and retry yourself with the same idempotency_key once a human confirms."New value: +"Create a durable owner-review request (default true). Setting false returns needs_approval, but an authority-envelope retry still cannot execute."

Changed21 schema fields changedJul 18, 2026

addedInput schema / properties / actor

Added value: +{
+  "description": "Who is asking: { type: agent|human|system|integration, id, display_name }.",
+  "type": "object"
+}

addedInput schema / properties / authority

Added value: +{
+  "description": "How the rollback was authorized: { mode, confirmed_by, confirmed_at, confirmation_ref }. human_confirmed requires all three confirmation fields; anything weaker returns pending_approval.",
+  "type": "object"
+}

addedInput schema / properties / business_id

Added value: +{
+  "description": "Business ID (optional only when the token can access exactly one business).",
+  "type": "string"
+}

addedInput schema / properties / dry_run

Added value: +{
+  "description": "Validate tenant, scope, and policy without touching the live config.",
+  "type": "boolean"
+}

addedInput schema / properties / idempotency_key

Added value: +{
+  "description": "Stable key for the rollback request; repeating it returns the original outcome. Defaults to a key derived from agent + version — pass a fresh key to repeat a rollback that already executed.",
+  "type": "string"
+}

addedInput schema / properties / queue_for_approval

Added value: +{
+  "description": "When authority is insufficient, create a durable dashboard approval (default true). Set false to get needs_approval and retry yourself with the same idempotency_key once a human confirms.",
+  "type": "boolean"
+}

addedInput schema / properties / source_ref

Added value: +{
+  "description": "External source reference, such as a ticket or automation run ID.",
+  "type": "string"
+}

addedOutput schema / properties / approval

Added value: +{
+  "type": [
+    "object",
+    "null"
+  ]
+}

addedOutput schema / properties / business_id

Added value: +{
+  "type": [
+    "string",
+    "null"
+  ]
+}

addedOutput schema / properties / code

Added value: +{
+  "type": [
+    "string",
+    "null"
+  ]
+}

addedOutput schema / properties / message

Added value: +{
+  "type": [
+    "string",
+    "null"
+  ]
+}

removedOutput schema / properties / new_version_number

Removed value: -{
-  "type": [
-    "integer",
-    "null"
-  ]
-}

addedOutput schema / properties / request_id

Added value: +{
+  "type": [
+    "string",
+    "null"
+  ]
+}

removedOutput schema / properties / restored_fields

Removed value: -{
-  "items": {
-    "type": "string"
-  },
-  "type": "array"
-}

removedOutput schema / properties / restored_from_version
```
Removed value: -{
-  "type": "integer"
-}
```

addedOutput schema / properties / result

Added value: +{
+  "description": "On executed: { agent_id, restored_from_version, restored_fields, skipped_fields, new_version_number, warning? }.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

addedOutput schema / properties / risk_level

Added value: +{
+  "type": [
+    "string",
+    "null"
+  ]
+}

removedOutput schema / properties / skipped_fields

Removed value: -{
-  "items": {
-    "type": "string"
-  },
-  "type": "array"
-}

addedOutput schema / properties / status
```
Added value: +{
+  "type": "string"
+}
```

addedOutput schema / properties / summary

Added value: +{
+  "type": [
+    "object",
+    "null"
+  ]
+}

removedOutput schema / properties / warning

Removed value: -{
-  "type": [
-    "string",
-    "null"
-  ]
-}

AddedJul 11, 2026

TDQS

A4.5/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

This is a destructive, idempotent, non-read-only tool, and the description adds meaningful behavioral context beyond the annotations: it changes live settings, requires authenticated owner review, explicitly states caller-supplied authority fields never authorize execution, and notes the executor preserves version recording and secret exclusion after approval. It also implies idempotent retry behavior consistent with the idempotentHint annotation.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Four sentences, each earning its place: what the tool does, what side effects and approvals apply, how to discover a target and complete the flow, and what the executor guarantees. The most decision-relevant facts are front-loaded, and the additional length is justified for a destructive, approval-gated operation.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the destructive annotations, a full input schema, an output schema, and 11 parameters, the description provides the missing operational context: where to find a valid target, the review URL relay step, and the auth model. An agent has enough information to select and correctly invoke the tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents all 11 parameters, including nuanced ones like idempotency_key, authority, and queue_for_approval. The tool description reinforces the authority limitation and the approval workflow, but it does not add parameter-level meaning beyond what the rich schema descriptions already provide.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and resource: 'Request a rollback to a prior assistant_config_versions snapshot.' It also names the live settings affected (greeting, voice, prompt, model), which clearly distinguishes it from generic agent configuration updates and from list-only tools like list_config_versions.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives a clear usage workflow: discover a target with list_config_versions or get_change_history, then relay the pending request's review URL. It explains the approval requirement and that authority fields cannot authorize execution. However, it does not explicitly state when to prefer rollback_config over the similarly adjacent update_agent_config sibling, leaving some inference to the agent.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All KaiCalls tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.