For the complete documentation index, see llms.txt.
001 hosted utility endpoints · since 2010

Every boring endpoint,
already built.

Calculators, converters, hashes, date math, and finance formulas as stable JSON APIs. One bearer token, one credit pool, schemas that don't drift. You ship the feature; we maintain the formula.

✓ OpenAPI 3.1 ✓ RFC 7807 errors ✓ MCP built in ✓ failed calls cost 0
Get a free key → View reference
✓ the same engines behind miniwebtool.com — tools people have searched for and used since 2010
LIVE 
$ curl https://api.miniwebtool.com/v1/tools/percentage-calculator/run \
    -H "Authorization: Bearer $MWT_KEY" \
    -d '{"mode":"percent_of","a":15,"b":200}'

{
  "request_id":   "01KPW8…",
  "tool":         "percentage-calculator",
  "credits_used": 1,
  "result": {
    "mode":   "percent_of",
    "a":      15,
    "b":      200,
    "result": 30
  }
}
import requests, os

r = requests.post(
    "https://api.miniwebtool.com/v1/tools/percentage-calculator/run",
    headers={"Authorization": f"Bearer {os.environ['MWT_KEY']}"},
    json={"mode": "percent_of", "a": 15, "b": 200},
    timeout=10,
)
print(r.json()["result"]["result"])  # 30
const res = await fetch(
  "https://api.miniwebtool.com/v1/tools/percentage-calculator/run",
  {
    method: "POST",
    headers: {
      "Authorization": `Bearer ${process.env.MWT_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({ mode: "percent_of", a: 15, b: 200 }),
  },
);
const { result } = await res.json();
console.log(result.result);  // 30
200 OK · 1 credit · application/json
Start free · no card
Paid plans from $29/mo for production volume
Endpoints
1060
curated, growing catalog
Auth
1 key
same bearer token everywhere
Cost per call
fixed
published credit weight per tool
Failed calls
0 credits
4xx/5xx auto-refund
agents start from a sentence, not a slug

Hand the whole catalog to your coding agent

You don't need to know which endpoint you want. Paste this into Claude Code, Cursor, Codex, Aider or Cline, then describe the problem in plain language — the agent searches the catalog, picks the tool, and wires it up without burning your key.

Discovery is free and needs no key: GET /v1/tools?q=mortgage+payment searches the whole catalog and returns slugs, summaries and credit costs. Only the final call is billed — and /dry-run costs 0.
Stable URL for this prompt: /prompt.md (?variant=mcp / ?variant=cursor) — point an agent straight at it.
Paste into any coding agent, or keep it in CLAUDE.md / AGENTS.md.
MiniWebtool is a growing catalog of deterministic calculation, conversion and text APIs behind one key — new tools land daily, so always search rather than assume. When I ask for something with an exact computed answer — a financial figure, a unit conversion, a hash, a statistic, a date — find the tool that does it instead of writing the formula yourself.

Search first, guess never:

1. `GET https://api.miniwebtool.com/v1/tools?q=<keywords>&limit=5` — public, no key. Search
   nouns (`mortgage`, `sha256`, `celsius`), not sentences. Results carry the
   slug, summary, and credit cost.
2. `GET https://api.miniwebtool.com/tools/<slug>.md` — that tool's fields and an example.
3. `POST https://api.miniwebtool.com/v1/tools/<slug>/dry-run` — validates your payload and
   returns a sample response for **0 credits**. Shape the request here.
4. `POST https://api.miniwebtool.com/v1/tools/<slug>/run` — the real call. Header:
   `Authorization: Bearer $MINIWEBTOOL_API_KEY`. Returns
   `{request_id, tool, credits_used, result}`.

Budget (1,000 free credits/month, most tools cost 1):

- Steps 1-3 are free. Iterate there.
- One live `/run` per tool while integrating. Never call it from tests or a
  retry loop — assert against the dry-run sample.
- 4xx: fix the payload, don't retry. 429: honour `Retry-After`.
- Stop and tell me if `X-MWT-Credits-Remaining` drops below 50.
- Tools are deterministic, so cache by input.

If `MINIWEBTOOL_API_KEY` isn't set, ask me — don't sign up or invent one.
One-time setup — the agent then sees every tool natively.
Connect the MiniWebtool MCP server, then use its tools for any exact
computation — a growing catalog behind one key.

```bash
claude mcp add --transport http miniwebtool https://api.miniwebtool.com/v1/mcp \
  --header "Authorization: Bearer $MINIWEBTOOL_API_KEY"
```

Or in an MCP config file:

```json
{
  "mcpServers": {
    "miniwebtool": {
      "type": "http",
      "url": "https://api.miniwebtool.com/v1/mcp",
      "headers": { "Authorization": "Bearer ${MINIWEBTOOL_API_KEY}" }
    }
  }
}
```

`tools/list` then gives you every tool with its JSON Schema — search it by name
or summary to match a plain-language need, and call it with typed arguments.

Calls are metered (1,000 free credits/month). Don't call tools from a
retry loop, and cache by input — results are deterministic.
Save as .cursor/rules/miniwebtool.mdc
---
description: MiniWebtool — growing catalog of calculation/conversion APIs
globs:
alwaysApply: false
---

MiniWebtool is a growing catalog of deterministic calculation, conversion and text APIs behind one key — new tools land daily, so always search rather than assume. When I ask for something with an exact computed answer — a financial figure, a unit conversion, a hash, a statistic, a date — find the tool that does it instead of writing the formula yourself.

Search first, guess never:

1. `GET https://api.miniwebtool.com/v1/tools?q=<keywords>&limit=5` — public, no key. Search
   nouns (`mortgage`, `sha256`, `celsius`), not sentences. Results carry the
   slug, summary, and credit cost.
2. `GET https://api.miniwebtool.com/tools/<slug>.md` — that tool's fields and an example.
3. `POST https://api.miniwebtool.com/v1/tools/<slug>/dry-run` — validates your payload and
   returns a sample response for **0 credits**. Shape the request here.
4. `POST https://api.miniwebtool.com/v1/tools/<slug>/run` — the real call. Header:
   `Authorization: Bearer $MINIWEBTOOL_API_KEY`. Returns
   `{request_id, tool, credits_used, result}`.

Budget (1,000 free credits/month, most tools cost 1):

- Steps 1-3 are free. Iterate there.
- One live `/run` per tool while integrating. Never call it from tests or a
  retry loop — assert against the dry-run sample.
- 4xx: fix the payload, don't retry. 429: honour `Retry-After`.
- Stop and tell me if `X-MWT-Credits-Remaining` drops below 50.
- Tools are deterministic, so cache by input.

If `MINIWEBTOOL_API_KEY` isn't set, ask me — don't sign up or invent one.
002 the contract

One envelope. Stable errors. Honest headers.

Everything you'd check before trusting a third-party API, in one glance. The same shape on every endpoint — integrate once, reuse for all 1060 tools.

response envelope
{
  "request_id":   "01KPW8…",
  "tool":         "<slug>",
  "tool_version": "1.0.0",
  "credits_used": 1,
  "result": { …typed output… }
}
errors · RFC 7807
{
  "status": 402,
  "code":   "insufficient_credits",
  "title":  "Insufficient credits",
  "request_id": "01KPW…"
}
// branch on `code` — it never changes
every response
X-RateLimit-Limit:       10
X-MWT-Credits-Used:      1
X-MWT-Credits-Remaining: 49998
X-Request-Id:            01KPW…
003 who builds on this
[saas]
Product teams
Calculators in dashboards, forms, and onboarding flows
[fin]
Financial sites
Interest, mortgage, APY, ROI — audited formulas
[edu]
Education apps
Practice, quizzes, step-by-step explanations
[mcp]
Agents & automation
Exact tool calls over MCP after the LLM plans
004 why pay for "simple"

The formula is the easy part.

Anyone can write a / b * 100. What eats your sprint is everything around it: input validation, edge cases, rounding rules, examples, docs that stay current, rate limits, and a billing story. We package that operational work behind one HTTP call — and we've been maintaining these exact tools in production since 2010.

Ship the feature this afternoon

POST a JSON body, render result. No formula review, no test matrix for edge cases you haven't thought of yet.

Integrate once, reuse 1060 times

Same auth, same envelope, same error codes on every endpoint. Your client wrapper is ~20 lines, total.

Predictable bill, no surprises

Each tool's credit weight is published in the OpenAPI spec. Failed calls refund automatically. Hard caps by default.

Best-fit use cases

  • Calculator features inside dashboards, forms, or onboarding flows.
  • Finance, education, and content sites that need utility endpoints that just work.
  • Agencies and automation builders running many small tools under one account.
  • AI agents that plan in natural language but need exact, deterministic execution.
Browse all endpoints → Read docs
005 pricing

Free to evaluate. Cheap to ship.

Start free, no card. Upgrade when an endpoint becomes part of your product. Cancel anytime from the Stripe portal — no sales call required.

Free

Free
1,000 credits/mo
1 req/sec
Hard cap
Testing and prototypes
Core text, hash, math, date, and conversion tools
Start free

Starter

$29/mo
75,000 credits/mo
5 req/sec
Hard cap + manual packs
Indie apps and automations
Core plus basic finance
Start Starter
POPULAR

Pro

$79/mo
400,000 credits/mo
15 req/sec
Hard cap default; optional $0.25 / 1K credits
Production SaaS and content workflows
Core, finance, text, random, and webmaster light
Start Pro

Business

$199/mo
1,500,000 credits/mo
50 req/sec
Hard cap default; optional $0.15 / 1K credits
Teams and production apps
All sync APIs, logs, invoices, team controls
Start Business

Scale

$499/mo
5,000,000 credits/mo
100 req/sec
Hard cap default; optional $0.10 / 1K credits
High throughput and batch usage
Higher limits, priority queue, crawler add-ons later
Start Scale

Yearly billing: pay 10 months, get 12. Need >5M/mo, SOC 2, or a DPA? [email protected]

First curl in 30 seconds.

Verify your email → get a key → stable JSON. No card, no SDK, no tool build-out.

Get a free key →