
AgentSLA MCP server
Which MCP servers are AgentSLA's own, from the official registry and from asking its own domain, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.
read 2026-11-16
- Registry nameio.agentsla/agentslaverified in the MCP registry
The registry name sits under agentsla.io. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.
- Stateanswerschecked 2026-11-17
- Tools10
- Server calls itselfAgentSLA
- Server version2.1.0
- Protocol version2025-06-18
On 2026-11-17 the server answered initialize and tools/list without a sign-in. First answer in 175 ms.
Should an agent connect it
- Whether its tools do the product's job has not been judged yet.
- Checked on 6 of the 7 days it takes to judge reliability.
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.
Measured daily
- INITE MCP scoreafter a weeka score needs 7 daily checks: 6 so far
- Answeredmeasuring6 of the 7 daily checks it takes to print a share
- Median first answer262 ms
- Space in an agent's context≈ 2,852 tokensthe definitions of all its tools (10), which an agent carries on every turn
- Tool definitionsA · 100 of 100
- Typed results0 of 10 toolsdeclared outputSchema: results come typed, read once
- Every tool passes all six checks.
Worth a look
None of the five warning signs in these definitions.
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 (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). 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 →
discover_monitored_apisSearches the 21,461 APIs and MCP servers indexed by AgentSLA by keyword and/or category. Use it first to find candidate services; then get_api_details for one listing, or check_endpoint to test a URL live. Free, read-only, served from the…
get_api_detailsReturns one directory listing in full: its id, name, endpoint_url, category, link_type, current_sla_score, any verified USDC rank stake, and a summary of its signed check history (number of checks, pass rate, latest check with latency)…
check_endpointProbes any HTTP(S) URL once, right now, from Cloudflare's edge and reports httpStatus, edgeLatencyMs and meetsSla (true when the status is 2xx/3xx within maxLatencyMs, or when an x402 API answers 402 with a valid payment request, whose…
verify_endpoint_slachanges dataProbes an endpoint (best of up to 3 tries) and returns an EIP-712 SLAAttestation signed by the AgentSLA oracle: httpStatus, latencyMs, slaMet, eip712Signature and oracleSigner, verifiable with ethers.verifyTypedData (domain and types: GET…
check_passReads from Base Mainnet whether a wallet holds an active 7-day pass (needed by verify_endpoint_sla), its expiry (validUntil, secondsLeft), the price (5 USDC), and the wallet's USDC balance and existing allowance to the escrow contract…
buy_passExplains how to buy a 7-day pass (5 USDC) for a wallet. Recommended: x402 at POST https://agentsla.io/v1/pass (one USDC signature, no gas, no approval). Alternative: the returned unsigned Base Mainnet transactions (a USDC approve, only if…
create_escrow_jobBuilds the unsigned Base Mainnet transactions that lock USDC in escrow for one directory listing: the provider gets the price if the oracle finds the endpoint meets your latency limit; otherwise you get the price plus the 2% fee back. Use…
register_escrow_termschanges dataRecords the latency limit for an escrow job after its createJob transaction confirms, so the oracle knows the SLA to judge. Use it once per job, right after the create_escrow_job transactions. The job is read from the transaction on Base…
get_escrow_jobReads one escrow job from Base Mainnet and AgentSLA's records: status (active, settled or reclaimed), creator, provider, listing and endpoint, price, fee, registered latency limit, created time, and when the creator may reclaim it if the…
get_usdc_escrow_contractReturns the reference data for AgentSLA's escrow contract on Base Mainnet: contract and USDC token addresses, chain id, the pass price and period, the 2% escrow fee, the 7-day reclaim timeout, the human-readable ABI of the functions…
How to connect
Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:
{
"mcpServers": {
"agentsla": {
"type": "http",
"url": "https://agentsla.io/mcp"
}
}
}Add it
Claude Code, in a terminal
claude mcp add --transport http agentsla https://agentsla.io/mcpclaude.ai, ChatGPT and other clients: add https://agentsla.io/mcp as a custom connector in the settings.
OpenAPI spec
13 paths, read 2026-11-16. 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.
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.