ExpenseBot

correct_expenses

Destructive

Safely correct the category, business purpose or explicit attendees on an exact bounded set of recorded expenses, e.g. 'categorize these as Travel', 'add Client kickoff dinner as the business purpose for these meals', 'add Fred, Rob, and Lamar to last night's business meal'. First use search_expenses to identify the exact rows, then pass their full expenseId values. A grounded preview is automatic: first call with confirm omitted/false, show the exact count and before-to-after changes, and ask once for approval. Only after explicit approval repeat the same operationId, selection and change with confirm:true (a premature confirm:true is converted to preview). Attendee names must come explicitly from the user; never infer them. attendeeMode add preserves existing attendees; replace substitutes only the attendee segment. Business purpose and attendees preserve the structured Notes field (card, description, inbox and other typed segments). Formula Notes and changes exceeding the Notes limit are skipped safely. The confirmed result reports applied/conflicted/failed counts and supports Undo. This tool does not omit duplicates, change amounts or dates, infer business context, or run broad Calendar matching. For unsupported or over-100-row cleanup, send the user to https://www.expensebot.ai/review-expenses?source=mcp.

Input Schema

Name	Required	Description
`undo`	No	Undo a completed correction within its Undo window.
`change`	No	Exactly one correction. The selected type determines which matching value field is required.
`status`	No	Read operation status using operationId.
`confirm`	No	Omit/false for preview; true only after explicit approval.
`selection`	No	Exact expenses returned by search_expenses and approved for this bounded correction.
`operationId`	Yes	Stable idempotency key generated once for preview and reused unchanged.

Output Schema

Name	Required	Description	Default
`undo`	No
`change`	No
`status`	Yes
`message`	No
`preview`	No
`success`	Yes
`manifest`	No
`finalAnswer`	No
`operationId`	No
`confirmation`	No
`responseGuidance`	No
`reviewExpensesUrl`	No
`requiresConfirmation`	No

Schema Changelog

Changes observed during successful MCP inspections.

Changed2 schema fields changedSep 9, 2026

addedInput schema / properties / change / oneOf

Added value: +[
+  {
+    "properties": {
+      "type": {
+        "const": "category"
+      }
+    },
+    "required": [
+      "category"
+    ]
+  },
+  {
+    "properties": {
+      "type": {
+        "const": "businessPurpose"
+      }
+    },
+    "required": [
+      "businessPurpose"
+    ]
+  },
+  {
+    "properties": {
+      "type": {
+        "const": "attendees"
+      }
+    },
+    "required": [
+      "attendees"
+    ]
+  }
+]

changedOutput schema / (root)

Previous value: -nullNew value: +{
+  "additionalProperties": true,
+  "properties": {
+    "change": {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "confirmation": {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "finalAnswer": {
+      "type": "string"
+    },
+    "manifest": {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "message": {
+      "type": "string"
+    },
+    "operationId": {
+      "type": "string"
+    },
+    "preview": {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "requiresConfirmation": {
+      "type": "boolean"
+    },
+    "responseGuidance": {
+      "type": "string"
+    },
+    "reviewExpensesUrl": {
+      "type": "string"
+    },
+    "status": {
+      "type": "string"
+    },
+    "success": {
+      "type": "boolean"
+    },
+    "undo": {
+      "additionalProperties": true,
+      "type": "object"
+    }
+  },
+  "required": [
+    "success",
+    "status"
+  ],
+  "type": "object"
+}

Changed1 schema field changedAug 25, 2026

changedOutput schema / (root)

Previous value: -{
-  "additionalProperties": true,
-  "description": "Standard ExpenseBot tool result envelope. `message` is the human-readable summary the AI cites; `data` is the structured payload (totals, breakdowns, ids, etc.). On failure, `success` is false and `error` carries a code/message/hint triple.",
-  "properties": {
-    "data": {
-      "additionalProperties": true,
-      "description": "Structured payload. Shape varies per tool — common keys: total, breakdown, comparison, sampleMeta, ids, expenseId, reportId, signupUrl, results.",
-      "type": "object"
-    },
-    "error": {
-      "additionalProperties": true,
-      "description": "Present only when success === false.",
-      "properties": {
-        "code": {
-          "type": "string"
-        },
-        "hint": {
-          "type": "string"
-        },
-        "message": {
-          "type": "string"
-        }
-      },
-      "type": "object"
-    },
-    "message": {
-      "description": "Human-readable result text. Always present on success; prefer rendering this verbatim before any further reasoning.",
-      "type": "string"
-    },
-    "sampleMeta": {
-      "additionalProperties": true,
-      "description": "Set when the underlying dataset was truncated. isTruncated=true means the agent saw a sample of `sampleCount` of `totalCount` rows; aggregate totals are still accurate.",
-      "properties": {
-        "isTruncated": {
-          "type": "boolean"
-        },
-        "sampleCount": {
-          "type": "integer"
-        },
-        "totalCount": {
-          "type": "integer"
-        }
-      },
-      "type": "object"
-    },
-    "success": {
-      "description": "False on tool errors; check before reading `data`.",
-      "type": "boolean"
-    }
-  },
-  "type": "object"
-}New value: +null

Changed8 schema fields changedAug 22, 2026

addedInput schema / properties / change / properties / attendees / description

Added value: +"Attendee names explicitly supplied by the user when type is attendees; never infer names."

addedInput schema / properties / change / properties / businessPurpose / description

Added value: +"User-supplied business purpose to apply when type is businessPurpose."

addedInput schema / properties / change / properties / category / description

Added value: +"Configured expense category to apply when type is category."

addedInput schema / properties / change / properties / type / description

Added value: +"The one correction kind to apply to every exact selected expense."

addedInput schema / properties / selection / description

Added value: +"Exact expenses returned by search_expenses and approved for this bounded correction."

addedInput schema / properties / selection / properties / items / description

Added value: +"One exact receipt identity per selected expense; never broaden this list after preview."

addedInput schema / properties / selection / properties / items / items / properties / expected / properties / category / description
```
Added value: +"Current category observed before preview for conflict detection."
```
addedInput schema / properties / selection / properties / items / items / properties / expected / properties / notes / description
```
Added value: +"Current structured Notes value observed before preview for conflict detection."
```

First observedAug 10, 2026

TDQS

A4.8/5.0

Behavior5/5

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

Annotations declare destructiveHint true and openWorldHint true; the description goes well beyond that by disclosing the mandatory preview gate, that a premature confirm:true is downgraded to preview, attendeeMode add-vs-replace semantics, that Notes structured segments are preserved while Formula Notes and over-limit changes are skipped, and that the confirmed result reports applied/conflicted/failed counts with Undo support. These are exactly the operational facts an agent needs before a destructive write.

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

Conciseness4/5

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

Front-loaded with purpose and examples, followed by the preview/confirm protocol and then constraints, so scanning order matches decision order. It is long and dense for a tool description, but nearly every sentence carries a rule an agent would otherwise get wrong; only the Notes/Formula detail is somewhat compressed.

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?

For a destructive, nested-object, 6-parameter tool with an output schema, the description supplies the full operating protocol (search → select → preview → approve → confirm → undo) plus the guardrails that annotations alone leave ambiguous. Return-value detail is correctly left to the output schema, with only the applied/conflicted/failed counts summarized.

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

Parameters4/5

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

Schema coverage is 100%, so 3 is the floor; the description still adds workflow meaning the schema cannot express, namely that operationId must be generated once and reused unchanged between preview and confirm, and that the selection list must not be broadened after preview. It also explains the conflict detection purpose of expected values indirectly through 'before-to-after changes'.

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?

States a concrete verb (correct) plus the exact resources it can touch (category, business purpose, attendees) on a bounded set of recorded expenses, and makes the scope explicit with 'exact bounded set' and an example set that maps directly to the three change types. It is clearly distinguishable from search_expenses (upstream selector) and update_expense (singular edit).

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

Usage Guidelines5/5

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

Explicitly prescribes the sequence: call search_expenses first, then pass full expenseId values, preview with confirm omitted/false, ask once, then confirm:true with the same operationId. It also states what the tool will not do (no duplicate omission, no amount/date edits, no inferred business context, no broad Calendar matching) and routes unsupported or over-100-row work to an external URL, giving an unambiguous alternative path.

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

All ExpenseBot tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.