About Mymemory
Mymemory is a Pipeworx translation MCP server that combines machine translation with translation-memory matches. It translates without an API key or registration, and returns match quality, detected source language when applicable, and alternative translations.
Key features
- 01Translate between languages
- 02Detect source language automatically
- 03Return match-quality score
- 04Provide up to three alternatives
- 05No translation API key required
Use cases
- 01Translate a message into another language
- 02Check whether a translation matches stored examples
- 03Compare alternate translations
- 04Translate without registering for a provider key
MCP support
The hosted endpoint exposes the Mymemory translation tool alongside shared Pipeworx meta-tools. A local npm stdio bridge forwards calls to that endpoint; meta-tools can be enabled separately in the bridge.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://gateway.pipeworx.io/mymemory/mcpclaude mcp add --transport http pipeworx-io-mymemory https://gateway.pipeworx.io/mymemory/mcp{"mcpServers":{"pipeworx-io-mymemory":{"type":"http","url":"https://gateway.pipeworx.io/mymemory/mcp"}}}Open in Cursor{"servers":{"pipeworx-io-mymemory":{"type":"http","url":"https://gateway.pipeworx.io/mymemory/mcp"}}}Open in VS Code{"mcpServers":{"pipeworx-io-mymemory":{"type":"http","url":"https://gateway.pipeworx.io/mymemory/mcp"}}}{"mcpServers":{"pipeworx-io-mymemory":{"serverUrl":"https://gateway.pipeworx.io/mymemory/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project README@pipeworx/mymemory
Keyless machine + translation-memory translation between languages — the one translation pack in the catalog that needs no API key, no self-hosted instance, and no registration of any kind.
Part of Pipeworx — an MCP gateway connecting AI agents to 1863+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.
Tools
mymemory_translate(text, target, source?)— translatestextintotarget(ISO 639-1, e.g."de").sourceis optional and defaults to automatic language detection; when it auto-detects, the detected code comes back indetected_source_language. Also returns a match-quality score (1.0 = an exact translation-memory match, lower = machine-translated) and up to 3 alternate candidate translations.
Auth
Keyless. Every call carries a courtesy contact address (not a login, not a key) that raises the shared anonymous allowance under the publisher's own usage policy — see "Quota" below.
Data sources
- MyMemory (Translated S.r.l.), a community and machine translation memory — translations contributed by professional translators, public corpora and machine translation, blended and ranked by match quality.
Quota
Free, anonymous use is limited by the publisher to 5,000 characters/day,
raised to 50,000 characters/day by supplying a contact address (which this
pack does on every call). All Pipeworx gateway traffic shares one egress
address, so this allowance is shared across every caller of this pack —
exactly like the SEC.gov egress note elsewhere in this repo. When the shared
daily allowance is spent, the tool returns a clear {found: false, reason: "rate_limit"} refusal rather than retrying or rotating identity; it
resets on the publisher's own rolling daily window, not ours.
Why this exists
deepl needs a funded platform key (unfunded by choice) and its shared key
also runs on a monthly character allowance that has sat exhausted for
extended stretches. translate and libretranslate both wrap LibreTranslate,
which needs either a key or a self-hosted endpoint since the free public
mirrors shut down in 2025-26. That left no translation tool anyone without a
key of their own could actually use — this pack is that tool (fleet #3862).
Both of those packs' refusals now name mymemory_translate as the keyless
alternative, and the gateway's sibling-retry (NAMED_SERVICE_SIBLING_OVERRIDE
in workers/gateway/src/retry-policy.ts) falls over to it automatically when
deepl_translate fails on a quota-exhausted shared key — the
source/source_lang and target/target_lang field-alias families in
shared/src/field-aliases.ts are what let the retry hand this pack the same
arguments the failed call received.
Egress
MyMemory's anonymous daily quota is counted per egress IP, and the shared Cloudflare range the gateway runs from is often already exhausted (HTTP 429). The pack tries the direct call first and, on a non-ok answer, retries once through Pipeworx's non-Cloudflare relay. If both are throttled, the tool returns an explicit rate_limit refusal rather than an error.
Quick Start
Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):
{
"mcpServers": {
"mymemory": {
"url": "https://gateway.pipeworx.io/mymemory/mcp"
}
}
}
What this endpoint actually serves
tools/list at https://gateway.pipeworx.io/mymemory/mcp returns the tools in the table
above plus the shared Pipeworx meta-tools — ask_pipeworx,
discover_tools, search_within, remember/recall and the rest of the
gateway-wide set. So the tool count you see is larger than this table: a
single-pack endpoint currently lists roughly 30 shared tools alongside the
pack's own. The connection's initialize response states its exact scope, and
is the authoritative answer for a given day.
This is deliberate, not multiplexing by accident. The meta-tools are what let a
scoped connection answer a question this pack does not cover — via
ask_pipeworx, which routes across the whole catalog — without you adding a
second MCP server. There is currently no way to mount a pack endpoint without
them; if the extra schemas cost you more context than the routing is worth,
connect to the full gateway once rather than to several pack endpoints.
Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:
{
"mcpServers": {
"pipeworx": {
"url": "https://gateway.pipeworx.io/mcp"
}
}
}
Both URLs reach the same gateway and the same 1863+ data sources. The
only difference is which pack's tools are listed directly; ask_pipeworx
reaches all of them from either one.
No MCP client? Call it over HTTP
curl -X POST https://gateway.pipeworx.io/v1/tools/mymemory_translate \
-H 'Content-Type: application/json' \
-d '{"text":"The meeting has been postponed until next week","target":"de"}'
No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/mymemory_translate. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.
Run it from npm
@pipeworx/mcp-mymemory is a local stdio bridge to the Pipeworx gateway: your MCP client
starts it with npx, and it forwards every call to the same
https://gateway.pipeworx.io/mymemory/mcp endpoint shown above. It needs network access;
there is no offline mode.
{
"mcpServers": {
"mymemory": {
"command": "npx",
"args": ["-y", "@pipeworx/mcp-mymemory"]
}
}
}
Without a key it calls anonymously, capped at 50 calls a day. A free account
raises that to 200 a day: get a key at https://pipeworx.io/signup and add
"env": { "PIPEWORX_API_KEY": "<your key>" } to the server entry.
The key is sent as a Bearer token, the same way a hosted connection sends it,
so calls are metered and billed exactly as they are there.
It lists only this pack's tools by default. Set PIPEWORX_META_TOOLS=1 to
also list ask_pipeworx and the other gateway-wide tools.
Behind an HTTPS proxy: Node's built-in fetch does not read HTTPS_PROXY by
default. Newer Node releases honour it when NODE_USE_ENV_PROXY=1 is set; check
your Node version's documentation.
Using with ask_pipeworx
Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:
ask_pipeworx({ question: "your question about Mymemory data" })
The gateway picks the right tool and fills the arguments automatically.
More
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.