monid

monid.

Name: oto
Author: Otomata

Find and read Monid endpoints — Monid is a paid gateway to ~2,000 data endpoints

ajouter à otoauth : —prefix : monid_*

setup & usage

⚠ your Monid API key, and the wallet it debits

create a key in the Monid dashboard → API keys (it starts with monid_), then paste it into oto.

each call is debited from the Monid wallet of the workspace the key is tied to: what you launch is paid there, not at oto

a key tied to no workspace answers 403; a revoked or badly copied key, 401 — create a new one and replace the old one on the card

the platform key opens only on explicit grant: without a grant, it is your key or your org's

⚠️ under the platform key, the Monid workspace is shared by the orgs that have a grant: the list of runs is refused there (it would show theirs); a run is re-read and stopped by its identifier

⚠️ under the platform key, the balance is refused too: it is that of the platform's shared wallet, not yours. A launch short of funds says so (402); to see a balance, set your own Monid key

the "test the connection" button calls the key's identity (free): it launches nothing and does not read the balance

▸ find the endpoint, read its price, launch it

Monid puts roughly 2,000 endpoints from about 70 providers behind a single key: web search, scraping, contact enrichment, social networks…

"find the work email of this LinkedIn profile" → monid_endpoint(q="work email from a linkedin profile") to find the endpoint, then monid_endpoint(op="inspect", provider=…, endpoint=…) for its input schema and price, finally monid_run(provider=…, endpoint=…, query_params={…})

long run (the launch returns a run not yet finished) → monid_runs(op="get", run_id=…, wait_seconds=30) until done: true

"stop that" → monid_runs(op="stop", run_id=…) (stops the spending; the stop is asynchronous)

"what have I launched?" → monid_runs() (newest to oldest) — with your own Monid key (or your org's): under the platform key, the list is refused

"how much is left?" → monid_wallet(): balance is the spendable amount (it can be negative), held what is reserved for runs in flight — with your own Monid key (or your org's): under the platform key, the balance is refused

· the input, the price and what is billed

a run's input has three parts — body, query_params, path_params — where inspect's schema places them, never flattened; provider and endpoint pass as discover returned them

the price that counts is the one from inspect. Its types are open (per call, per result with a flat fee, per unit, tiered, per matrix…), and volume parameters (maxItems, limit) often apply per query: three terms × 10 results is 30 billed results

⚠️ for a tiered or matrix price, amount is only a base, sometimes 0 on a paid endpoint: the real rate is in default / tiers / variants, and a tier read on the output is only known after the run, so the displayed price is a floor

⚠️ run status ≠ provider status: COMPLETED means "the provider answered", whatever it answered. COMPLETED + 404 = "nothing found", not billed; provider_ok settles it for you — and stays empty when Monid gives no provider status (then read next_step)

BLOCKED = a Monid workspace cap (budget, number of runs) refused the run before execution: nothing is billed, and relaunching blocks again until the cap is changed at Monid. FAILED (failure on Monid's side), TIMED_OUT and STOPPED are not billed either

⚠️ never relaunch a run whose outcome is unknown: Monid has no idempotency key, relaunching can pay twice. That is the case when the launch does not answer in time — a synchronous provider holding the connection longer than ~35 s returns this refusal, and the run probably exists. Finding it in monid_runs() requires your own Monid key; under the platform key, an administrator looks at the platform's workspace

the wait on a launch is bounded (40 s at most, ~55 s in per-socket timeouts — as long as the connection is established on the first try: each unreachable address adds up to 10 s, and DNS resolution is not bounded): beyond that, the tool returns the run in progress and the next step rather than making the client hang up

cost_usd is what Monid declares billed, read from the response and never recomputed; it stays empty until the run is settled

· verification status

written against the OpenAPI contract 0.1.0 published by Monid, and tested on a fake transport that replays its responses (runs returned under an HTTP code copying the provider's, error envelope, 202 then re-read, stop 409, wallet). Not yet exercised against a real account: neither the exact shape of whoami, nor the cursor of the list of runs, nor the real format of the key.

out of reach here, on purpose: the workspace's budgets and run caps, resources, API key management, wallet top-up and history, the public registry. Monid's hints are not requested (no client header is sent) and are not read.

⚙ multiple comptes

this connector accepts multiple comptes: each stored credential becomes a named account (one name per compte), at your level, your team's or your org's.

the first one doesn't need a name; from the second on, each one carries its own (the first then takes the name "principal", renamable)

without specification, the agent takes the only stored compte, otherwise the one marked as default; otherwise the call is refused, naming the available comptes — never a random choice

target a compte for a call: _account="<name>" on the tool; list them: oto_identity(op='list', connector='monid') (scope='org' or scope='group' for the org's or team's)

set the default: oto_identity(op='set', connector='monid', identity_id='<name>'); rename: op='rename' with new_name

⚠️ no tool walks through the comptes on its own: a total across several comptes is obtained by calling each one (_account) and adding up

outils

chargement…

utilisation

claude code — installe le plugin oto — claude plugin marketplace add otomata-tech/oto-plugin — mcp + skill configurés.

claude.ai — branche mcp.oto.cx/mcp en connecteur custom (guide)

cli — pipx install oto-cli puis oto monid …

← tous les connecteurs catalogue mcp →