
Swarmio MCP server
Which servers in the official MCP registry are Swarmio's own, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.
read 2026-10-26
- Registry nameio.ioswarm/swarmioverified in the MCP registry
The registry name sits under ioswarm.io. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.
- Stateanswerschecked 2026-10-26
- Tools14
- Server calls itselfswarmio
- Server version0.1.0
- Protocol version2025-06-18
On 2026-10-26 the server answered initialize and tools/list without a sign-in. First answer in 70 ms.
- Its tools do every use case we checked: compile cited reports on open questions from one request; compress the day's news into sourced briefs.
- Checked on 1 of the 7 days it takes to judge reliability.
Once connected
- runs.cancel change data without a sign-in: anyone with the address can call them.
Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.
- INITE MCP scoreafter a weeka score needs 7 daily checks: 1 so far
- What an agent can do with it2 of 2 use cases
- Answeredmeasuring1 of the 7 daily checks it takes to print a share
- Median first answer208 ms
- Space in an agent's context≈ 1,552 tokensthe definitions of all its tools (14), which an agent carries on every turn
- Tool definitionsA · 99 of 100
- Typed results14 of 14 toolsdeclared outputSchema: results come typed, read once
- No description an agent can choose by: 1 of 14
What an agent can do with it
- can do · compile cited reports on open questions from one request
- can do · compress the day's news into sourced briefs
- not sure · compare product prices and trade-offs with sources
Each of the product's use cases, put to Jev against this server's tool list: can an agent do it with these tools?
Worth a look
- Tools that change data, on a server that answers without a sign-in: runs.cancel
A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →
market.pulseFREE, no key, no account: live prices for BTC, ETH, SOL, AMZN and NVDA with the session's change, plus today's news — the platform's newest published news digest parsed into stories with source links, and the market wire's latest…
searchSearch the swarm's published research: the featured shelf, the operator's daily news / markets / AI digests, and — when a key is presented — the caller's own reports. Every hit is a finished, cited, evidence-graded report. Returns…
fetchRead one report by the id search returned: the full markdown text with its citations and evidence header, plus title, url and metadata. Public reports need no key; a caller's own reports need theirs.
web.searchFREE, no key, no account: LIVE web search, 25 queries a day per caller, up to 5 results each ({results:[{url,title,snippet,published}], quota, report}). This is the open web, not the published corpus — for the platform's finished reports…
news.briefPAID, 5¢ per call over x402 (USDC on Base, no account): today's market-news brief — this platform's newest published news digest synthesized with the live market wire into a cited bullet list, every finding ending with its source URL. No…
route.classifyClassify a freeform request into the right mode (research, news, finance, astrology, …) before spending a run. Never charged.
runs.launchLaunch a run: agents search the web, read sources, and write a cited report — real work, done for the caller. BILLED from the caller's credits. Returns the run record; poll runs.get until completed, then runs.report. Modes: research…
runs.getA run's status and metadata. Poll until status is completed/failed/cancelled.
runs.waitBlock until a run finishes, then return its record — and its report when it completed. Runs take 3–10 minutes; call this instead of polling runs.get in a loop. Returns early with timed_out=true after timeout_secs (default 90, max 240)…
runs.reportThe finished, cited report for a completed run (report_text is markdown).
runs.cancelStop a running job.
billing.balanceThe caller's credit balance and plan state.
billing.topup_termsHow to refill credits machine-to-machine: USDC on Base to a published address, verified on-chain. Returns the pay_to address, minimum, and rate. Public — no credentials needed.
modes.listThe launchable modes with a one-line contract for each. Static; never charged, no credentials needed.
How to connect
Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:
{
"mcpServers": {
"swarmio": {
"type": "http",
"url": "https://ioswarm.io/mcp"
}
}
}Add it
Claude Code, in a terminal
claude mcp add --transport http ioswarm https://ioswarm.io/mcpclaude.ai, ChatGPT and other clients: add https://ioswarm.io/mcp as a custom connector in the settings.
OpenAPI spec
45 paths, read 2026-10-26. Open the spec ↗
Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.
Cookies
Before we count this visit
Analytics and ad measurement tell us which pages and which of our ads do their job: Google Analytics and Google Ads, the Meta pixel, Yandex Metrica, and one first-party id of our own. Nothing is stored until you answer, and declining changes nothing about the site.