toolforte

Exact IBAN, VAT, cron, regex answers; HTML/URL to hosted PDF or screenshot; agent memory; workflows.

我該用這個嗎

品質與安全性

B
說明品質
95%
結構描述完整度
99%
命名品質
89%
汙染風險
40%
權限相符程度
100%
協定合規性
100%

發現項目(5)

  • HIGHTool poisoning patterns detected
  • MEDIUMTool description contains URL to non-standard domain在 workflow_list 中
  • MEDIUMTool description contains URL to non-standard domain在 workflow_propose 中
  • MEDIUMTool description contains URL to non-standard domain在 workflow_run 中
  • LOWTool 'dutch_public_holidays' description lacks action verb在 dutch_public_holidays 中

根據工具定義與協定合規性的自動化分析。

上下文成本

~4,036Token(工具定義)
~2.0 KB典型回應大小
顯著的注意力影響(128k 上下文的 3.15%)

這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高,可用於其他工作的注意力就越少。

安裝

一鍵安裝

將以下內容加入你的 `claude_desktop_config.json` 檔案:

{
  "mcpServers": {
    "toolforte": {
      "command": "npx",
      "args": [
        "toolforte-mcp"
      ]
    }
  }
}

可執行的套件

npmtoolforte-mcp0.3.0stdio

遠端端點

https://toolforte.com/api/mcp/mcpstreamable-http
https://toolforte.com/api/mcpstreamable-http

它能做什麼

工具清單

工具(14)

🟢 唯讀🟡 寫入🔴 刪除⚪ 未知
🟢text_diff(original, modified)

Compute an exact line-based diff between two texts with add/remove/unchanged stats.

輸入結構描述

{
  "type": "object",
  "properties": {
    "original": {
      "type": "string",
      "maxLength": 100000,
      "description": "The text before the change"
    },
    "modified": {
      "type": "string",
      "maxLength": 100000,
      "description": "The text after the change"
    }
  },
  "required": [
    "original",
    "modified"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "lines": {
      "type": "array",
      "items": {}
    },
    "stats": {
      "type": "object",
      "properties": {
        "added": {
          "type": "number"
        },
        "removed": {
          "type": "number"
        },
        "unchanged": {
          "type": "number"
        }
      },
      "additionalProperties": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢format_json(json, indent)

Validate and pretty-print JSON with a real parser. Returns formatted JSON or the exact parse error.

輸入結構描述

{
  "type": "object",
  "properties": {
    "json": {
      "type": "string",
      "maxLength": 200000,
      "description": "The JSON text to validate and format"
    },
    "indent": {
      "default": 2,
      "description": "Spaces per indentation level, 0 to 8 (default 2)",
      "type": "integer",
      "minimum": 0,
      "maximum": 8
    }
  },
  "required": [
    "json"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "output": {
      "description": "The final output of the workflow",
      "type": "string"
    },
    "valid": {
      "description": "Whether the input passed every check",
      "type": "boolean"
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢csv_to_json(csv, firstRowHeaders)

Parse CSV data into JSON with delimiter auto-detection and quoted-field handling.

輸入結構描述

{
  "type": "object",
  "properties": {
    "csv": {
      "type": "string",
      "maxLength": 200000,
      "description": "The CSV text; the delimiter is detected automatically"
    },
    "firstRowHeaders": {
      "default": true,
      "description": "Treat the first row as column names and return objects (default true)",
      "type": "boolean"
    }
  },
  "required": [
    "csv"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "json": {
      "type": "array",
      "items": {}
    },
    "delimiter": {
      "type": "string"
    },
    "row_count": {
      "type": "number"
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢test_regex(pattern, text, flags)

Run a JavaScript regular expression against a test string using a real regex engine. Returns all matches with capture groups and indices.

輸入結構描述

{
  "type": "object",
  "properties": {
    "pattern": {
      "type": "string",
      "maxLength": 1000,
      "description": "JavaScript regular expression source, without the slashes"
    },
    "text": {
      "type": "string",
      "maxLength": 20000,
      "description": "The text to run the expression against"
    },
    "flags": {
      "default": "g",
      "description": "Regex flags such as g, i, m, s, u (default g)",
      "type": "string",
      "maxLength": 8
    }
  },
  "required": [
    "pattern",
    "text"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "matches": {
      "type": "array",
      "items": {}
    },
    "count": {
      "description": "Number of items",
      "type": "number"
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢validate_iban(iban)

Check whether an IBAN is well formed: the mod-97 checksum, the country code, the length that country's IBAN must have, and whether every position holds the kind of character that country allows there (mod-97 turns letters into numbers, so on its own it cannot see a capital O typed in place of a zero). Returns validity, country, and where known the bank identifier printed inside the IBAN. This is a format check, not a bank check: a passing IBAN can still name a bank code that never existed or has since merged away, and says nothing about whether the account exists or can receive a payment. For twelve countries it also verifies the national check digit that the bank computes over the bank code and account number (BE, FR, MC, ES, IT, SM, PT, SI, PL, TN, EE, HU); nationalCheck 'failed' there means the number is very probably invented, since that digit cannot be back-computed the way the mod-97 pair can. Absent nationalCheck means not checked, never approved. For Dutch IBANs it also names the institution behind the four-letter bank code, including codes that no longer issue accounts; an unrecognised code does not make the IBAN invalid. A territory that borrows another country's IBAN (Reunion, Guernsey, Aland and the like) is reported with the prefix it actually uses, since no IBAN starts with RE or GG. Never guess whether an IBAN is well formed: use this tool. Do not present a pass as confirmation that money can be sent.

輸入結構描述

{
  "type": "object",
  "properties": {
    "iban": {
      "type": "string",
      "maxLength": 64,
      "description": "The IBAN to validate"
    }
  },
  "required": [
    "iban"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "valid": {
      "description": "Whether the input passed every check",
      "type": "boolean"
    },
    "iban": {
      "type": "string"
    },
    "country": {
      "type": "string"
    },
    "countryName": {
      "type": "string"
    },
    "checkDigits": {
      "type": "string"
    },
    "bankCode": {
      "type": "string"
    },
    "accountNumber": {
      "type": "string"
    },
    "institution": {
      "type": "object",
      "properties": {
        "name": {
          "type": "string"
        },
        "kind": {
          "type": "string"
        }
      },
      "additionalProperties": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢calculate_vat(amount, rate, mode)

Add VAT to a net amount or extract VAT from a gross amount at any rate (e.g. 21 for Dutch BTW). Returns exact net, VAT, and gross amounts rounded to cents.

輸入結構描述

{
  "type": "object",
  "properties": {
    "amount": {
      "type": "number",
      "minimum": 0,
      "description": "The amount"
    },
    "rate": {
      "type": "number",
      "minimum": 0,
      "maximum": 100,
      "description": "VAT rate in percent"
    },
    "mode": {
      "type": "string",
      "enum": [
        "add",
        "extract"
      ],
      "description": "add = amount is net, add VAT on top; extract = amount is gross, split out VAT"
    }
  },
  "required": [
    "amount",
    "rate",
    "mode"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "net": {
      "type": "number"
    },
    "vat": {
      "type": "number"
    },
    "gross": {
      "type": "number"
    },
    "rate": {
      "type": "number"
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢dutch_public_holidays(year)

Exact Dutch public holidays for a year, including movable feasts (Easter, Ascension, Whitsun) and King's Day rules. Use this instead of computing Dutch holidays yourself.

輸入結構描述

{
  "type": "object",
  "properties": {
    "year": {
      "type": "integer",
      "minimum": 1900,
      "maximum": 2200,
      "description": "Calendar year, 1900 to 2200"
    }
  },
  "required": [
    "year"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "year": {
      "type": "number"
    },
    "holidays": {
      "type": "array",
      "items": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢working_days_between(start, end, excludeDutchHolidays)

Count working days (Mon-Fri) between two dates inclusive, optionally excluding Dutch public holidays. Exact date math for SLA, payroll, and planning questions.

輸入結構描述

{
  "type": "object",
  "properties": {
    "start": {
      "type": "string",
      "pattern": "^\\d{4}-\\d{2}-\\d{2}$",
      "description": "YYYY-MM-DD"
    },
    "end": {
      "type": "string",
      "pattern": "^\\d{4}-\\d{2}-\\d{2}$",
      "description": "YYYY-MM-DD"
    },
    "excludeDutchHolidays": {
      "default": true,
      "description": "Skip Dutch public holidays that fall on a weekday (default true)",
      "type": "boolean"
    }
  },
  "required": [
    "start",
    "end"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "start": {
      "type": "string"
    },
    "end": {
      "type": "string"
    },
    "totalDays": {
      "type": "number"
    },
    "workingDays": {
      "type": "number"
    },
    "weekendDays": {
      "type": "number"
    },
    "holidaysExcluded": {
      "type": "array",
      "items": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢workflow_list(limit)

List the workflows saved to the ToolForte account this API key belongs to. A workflow runs several ToolForte tools in order, passing each one the output of an earlier step, so it does a whole job in one call instead of you orchestrating the steps yourself. Returns each workflow's id, the inputs it expects, the steps it runs and what a run costs. Call this first: workflow_run needs an id from here. Requires an API key, because a workflow belongs to an account. Get a free key at https://toolforte.com/developers and send it on the MCP connection as the header Authorization: Bearer tf_...

輸入結構描述

{
  "type": "object",
  "properties": {
    "limit": {
      "default": 50,
      "description": "How many to return, newest first.",
      "type": "integer",
      "minimum": 1,
      "maximum": 200
    }
  }
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "count": {
      "description": "Number of items",
      "type": "number"
    },
    "workflows": {
      "description": "The workflows on the account",
      "type": "array",
      "items": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟡workflow_propose(goal, save)

Describe a job in plain words and get back a workflow built only from tools this server really has: a name, the inputs it needs, and the steps in order, each with one sentence on why it is there. The proposal is honest about gaps: partial is true and missing names what could not be done, and when nothing here fits the answer says so instead of inventing a step. Send save: true to keep it on the account, after which workflow_run can run it by the returned id. Costs 10 credits, the price of one AI generation, charged once per proposal. Requires an API key linked to a ToolForte account. Get a free key at https://toolforte.com/developers and send it on the MCP connection as the header Authorization: Bearer tf_...

輸入結構描述

{
  "type": "object",
  "properties": {
    "goal": {
      "type": "string",
      "minLength": 8,
      "maxLength": 1000,
      "description": "The job to be done, in a sentence or two, for example: take a CSV of invoices, sum the amounts per customer and write the totals to a spreadsheet."
    },
    "save": {
      "default": false,
      "description": "Save the proposed workflow to the account so workflow_run can run it. Default false: the proposal is returned for you to read first.",
      "type": "boolean"
    }
  },
  "required": [
    "goal"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "definition": {
      "description": "The proposed workflow: name, declared inputs and steps",
      "type": "object",
      "properties": {},
      "additionalProperties": {}
    },
    "explanation": {
      "description": "Plain words on what the workflow does and why each step is there",
      "type": "string"
    },
    "steps": {
      "description": "What each step produced",
      "type": "array",
      "items": {}
    },
    "partial": {
      "description": "Whether the workflow only does part of what was asked",
      "type": "boolean"
    },
    "missing": {
      "description": "What could not be done, named",
      "type": "array",
      "items": {}
    },
    "creditsCharged": {
      "description": "Credits spent by this call",
      "type": "number"
    },
    "saved": {
      "description": "The saved workflow's id and name, when save was true",
      "type": "object",
      "properties": {
        "workflowId": {
          "type": "string"
        },
        "name": {
          "type": "string"
        }
      },
      "additionalProperties": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟡workflow_run(workflowId, inputs)

Run one saved workflow by id and return what every step produced. Get the id and the input keys from workflow_list. Each step costs the credits its own tool costs, normally 1 per step, and the credit for this call counts as the first step's rather than being charged on top. Steps run in order and stop at the first failure; the result says which step stopped it and what was spent up to that point. Check the "substitutions" field when it is present: it lists values a step supplied that the tool replaced with its own, which means the run succeeded on a value you did not send. Requires an API key, because a workflow belongs to an account. Get a free key at https://toolforte.com/developers and send it on the MCP connection as the header Authorization: Bearer tf_...

輸入結構描述

{
  "type": "object",
  "properties": {
    "workflowId": {
      "type": "string",
      "maxLength": 100,
      "description": "The id of a workflow from workflow_list."
    },
    "inputs": {
      "description": "Values for the workflow's declared inputs, keyed by the input key that workflow_list reports. Omit for a workflow that declares none.",
      "type": "object",
      "propertyNames": {
        "type": "string"
      },
      "additionalProperties": {}
    }
  },
  "required": [
    "workflowId"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "workflowId": {
      "type": "string"
    },
    "name": {
      "type": "string"
    },
    "ok": {
      "description": "Whether the run completed",
      "type": "boolean"
    },
    "runId": {
      "type": "string"
    },
    "creditsCharged": {
      "description": "Credits spent by this call",
      "type": "number"
    },
    "durationMs": {
      "type": "number"
    },
    "stepsRun": {
      "type": "number"
    },
    "stepsDefined": {
      "type": "number"
    },
    "output": {
      "description": "The final output of the workflow",
      "type": "object",
      "properties": {},
      "additionalProperties": {}
    },
    "steps": {
      "description": "What each step produced",
      "type": "array",
      "items": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢search_tools(query, limit)

Find ToolForte tools by what you want done, for example "iban", "vat", "pdf merge", "working days", "csv to json", "screenshot". ToolForte has over a hundred exact tools (validation, conversion, Dutch tax and dates, test data, files, AI); a session lists a small core of them and this is how you reach the rest. Returns the best matches with their arguments and cost. Then call the tool by name, or through run_tool. Free.

輸入結構描述

{
  "type": "object",
  "properties": {
    "query": {
      "default": "",
      "description": "Words describing the job. Empty lists every tool the server has.",
      "type": "string",
      "maxLength": 200
    },
    "limit": {
      "default": 8,
      "description": "How many matches to return, 1 to 50 (default 8)",
      "type": "integer",
      "minimum": 1,
      "maximum": 50
    }
  }
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string"
    },
    "tools": {
      "type": "array",
      "items": {}
    },
    "count": {
      "description": "Number of items",
      "type": "number"
    },
    "total": {
      "type": "number"
    },
    "note": {
      "type": "string"
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢describe_tool(name)

The full definition of one ToolForte tool by name: description, argument schema (JSON Schema), output schema, annotations, cost and an example call. Use it after search_tools when the argument summary is not enough. Free.

輸入結構描述

{
  "type": "object",
  "properties": {
    "name": {
      "type": "string",
      "maxLength": 80,
      "description": "The tool name, for example validate_iban"
    }
  },
  "required": [
    "name"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "name": {
      "type": "string"
    },
    "title": {
      "description": "Page title",
      "type": "string"
    },
    "description": {
      "type": "string"
    },
    "credits": {
      "type": "number"
    },
    "listed": {
      "type": "boolean"
    },
    "inputSchema": {
      "type": "object",
      "properties": {
        "type": {
          "type": "string"
        },
        "properties": {
          "type": "object",
          "properties": {},
          "additionalProperties": {}
        }
      },
      "additionalProperties": {}
    },
    "outputSchema": {
      "type": "object",
      "properties": {
        "type": {
          "type": "string"
        }
      },
      "additionalProperties": {}
    },
    "annotations": {
      "type": "object",
      "properties": {
        "readOnlyHint": {
          "type": "boolean"
        }
      },
      "additionalProperties": {}
    },
    "example": {
      "type": "object",
      "properties": {
        "name": {
          "type": "string"
        },
        "arguments": {
          "type": "object",
          "properties": {
            "iban": {
              "type": "string"
            }
          },
          "additionalProperties": {}
        }
      },
      "additionalProperties": {}
    },
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}
🟢run_tool(name, arguments)

Run any ToolForte tool by name with its arguments, whether or not it is listed in this session. The result, the cost and the metering are exactly what the tool gives when called directly; run_tool itself is free. Find names with search_tools and schemas with describe_tool.

輸入結構描述

{
  "type": "object",
  "properties": {
    "name": {
      "type": "string",
      "maxLength": 80,
      "description": "The tool name, for example calculate_vat"
    },
    "arguments": {
      "default": {},
      "description": "The tool's arguments, as describe_tool documents them",
      "type": "object",
      "propertyNames": {
        "type": "string"
      },
      "additionalProperties": {}
    }
  },
  "required": [
    "name"
  ]
}

輸出結構描述

{
  "type": "object",
  "properties": {
    "result": {
      "description": "The result, when it is not an object"
    }
  },
  "additionalProperties": {}
}

社群

為此伺服器評分

證據

近期觀測

已驗證未記錄版本14 個工具
已驗證未記錄版本14 個工具
已驗證未記錄版本14 個工具
已驗證未記錄版本14 個工具