Math Tool — Deterministic Math Calculator for AI Agents

Math Tool is a lightweight Apify Actor that lets AI agents, LLM workflows, and automations evaluate mathematical expressions with deterministic, reproducible results. Offload arithmetic, powers, roots, and trigonometry from your agent to a dedicated calculator instead of relying on probabilistic model output.

Run it from Apify Console, the Apify API, scheduled jobs, or agent frameworks that call external tools. Results are returned as structured JSON you can pipe into the next step of any workflow.

What is Math Tool?

Math Tool evaluates one or more math expressions in a single run and returns exact numeric answers. It uses the expr-eval parser — the same engine behind many agent math utilities — so expressions like 2 + 2, sqrt(16), sin(pi / 2), and 3^4 are parsed and calculated server-side, not guessed by an LLM.

Ideal for: AI agents, RAG pipelines, finance automations, engineering calculators, data validation, and any workflow that needs reliable math over the Apify API.

What can this Math Tool do?

• Evaluate multiple expressions in one API call
• Return one JSON dataset row per expression with expression, result, and error
• Support arithmetic, exponents, roots, trigonometry, logarithms, and variables
• Run in milliseconds on minimal compute — no browser, no proxy, no scraping
• Integrate via Apify API, webhooks, schedules, and third-party automation platforms
• Export results as JSON, CSV, or Excel from the dataset

Why use Math Tool for AI agents?

Large language models often hallucinate numbers. Giving your agent a deterministic math tool on Apify means:

1. Accuracy — Calculations are computed, not inferred.
2. Reproducibility — The same expression always yields the same result.
3. Auditability — Every run stores input expressions and outputs in a dataset.
4. Scalability — Call the Actor from thousands of agent runs via API without hosting your own calculator.

Pair Math Tool with other Apify Actors to build agent workflows: scrape data, calculate metrics, and push results anywhere.

How do I use Math Tool from an AI agent?

1. Open Math Tool on Apify Store or run it from Apify Console.
2. Pass an array of expressions in the input.
3. Read one dataset row per expression from the run output or via the Apify API.

API example

{
  "expressions": ["2 + 2", "sqrt(16)", "(10 + 5) * 3", "log(100)"]
}

Output example

Each expression produces one dataset row:

[
  { "expression": "2 + 2", "result": 4, "error": null },
  { "expression": "sqrt(16)", "result": 4, "error": null },
  { "expression": "(10 + 5) * 3", "result": 45, "error": null },
  { "expression": "log(100)", "result": 4.605170185988092, "error": null }
]

Download the dataset in JSON, CSV, HTML, or Excel from the run overview.

Supported expression syntax

Math Tool uses expr-eval. Common operators and functions:

Category	Examples
Arithmetic	+, -, *, /, %, ^
Grouping	(, )
Constants	PI, E, pi, e, true, false
Roots & powers	sqrt(x), cbrt(x), x^y
Trigonometry	sin(x), cos(x), tan(x), asin(x), acos(x), atan(x)
Logarithms	log(x), ln(x), log10(x)
Rounding	abs(x), ceil(x), floor(x), round(x)
Variables	x = 5; x * 2 (assign then use in the same expression string)

Full parser documentation: expr-eval on GitHub.

Input

Math Tool has the following input options. Click the Input tab in Apify Console for field descriptions.

Field	Type	Description
expressions	string[]	Required. One or more math expressions to evaluate.

How much does Math Tool cost?

Math Tool is extremely lightweight — no browser, no proxies, and typically finishes in under one second. A run consumes a fraction of a Compute Unit, so you can evaluate thousands of expressions on Apify's free tier. Exact cost depends on your Apify plan and run frequency.

FAQ

Is Math Tool safe for agent tool calling?

Yes. Expressions are parsed and evaluated by expr-eval in an isolated Apify container. There is no arbitrary code execution — only supported math syntax is accepted.

Can I call Math Tool from Python or JavaScript agents?

Yes. Use the Apify API or official Apify client libraries to start a run, pass expressions, and read results from the default dataset.

What happens when an expression fails?

Failed expressions do not stop the run. Each row includes result: null and an error message. Other expressions in the same batch are still evaluated.

[
  { "expression": "2 + 2", "result": 4, "error": null },
  {
    "expression": "foo(bar)",
    "result": null,
    "error": "undefined variable: foo"
  }
]

Why not let the LLM calculate directly?

LLMs approximate math and can be wrong on large numbers, floating-point edge cases, or multi-step formulas. Math Tool gives agents a reliable calculator API on Apify.

Integrations

• Apify API — Start runs and fetch dataset items programmatically
• Webhooks — Trigger downstream steps when evaluation completes
• Schedules — Recompute metrics on a cron
• Make, Zapier, n8n — Connect agent workflows without custom hosting

Support

Open an issue on the Actor page if you need additional functions. Math Tool is designed to stay minimal and fast for agent use cases.