MCP Gateway vs Monid

MCP Gateway 30

Self-hosted rule, scored on the paid option because any company use needs a commercial licence, as gaia and screenpipe (also noncommercial licences) were read. No payment protocol (0). The standard commercial licence is a public flat EUR 500 a month per named project through GitHub Sponsors, plan-only pricing (10). Personal and noncommercial use is free with no card or account, but there is no free route or stated evaluation period for a company (10 of 20). It installs from Homebrew, crates.io or ghcr.io with no sign-up, and an agent with a shell can run the setup wizard, but buying the licence takes a person on GitHub Sponsors, as screenpipe's did (10 of 20). 0 + 10 + 10 + 10 = 30.

Monid 85

x402 on Monid's own run route, POST https://x402.monid.ai/v1/run, in USDC on Base or Monad through a facilitator with no account, and run retrieval by a Sign-In-With-X signature from the paying wallet, per the 'Pay with x402' guide and the OpenAPI file's error codes. Metered endpoints and those tied to workspace resources or with an unbounded price are excluded, the floor is $0.01, and discovery and inspection need a key. Exa took 30 for x402 on its main paid routes with its other products excluded, the reading taken here (30 of 40). Not probed, because robots.txt on x402.monid.ai disallows every path. Per-call and per-result prices on public pages without a login, such as $0.00015 a call for Litescrape's Google Maps search, and discover and inspect return each price before a run (20). New accounts start with $1 of free credit and no card, per Monid (20). An agent with a funded wallet can run and fetch over x402 with no account, but the documented loop needs an API key, which a person creates in the dashboard, for discovery and input schemas, and topping up a balance needs a person to pay a Stripe checkout. Firecrawl took 15 when part of its surface needed a key a person creates (15 of 20).

MCP Gateway 87

Local-package reading, as for graphicle and claude-mem, since the owner runs it. Official packages on crates.io (cargo install mcp-gateway), a Homebrew tap, release binaries and ghcr.io, with Rust 1.95 and macOS, Linux and Windows builds stated (20). A public CI workflow runs clippy with warnings as errors, the full test suite, fmt, cargo audit, secret scans and Helm smoke tests on pushes and pull requests, and the last eight runs on main passed by 26 October 2026 (25). Four open issues on GitHub on 2 November 2026, one a labelled bug filed by the maintainer, and the newest outside issue was answered the day it was opened. Closed issues couldn't be read (robots.txt), so the record is partial (18 of 25). Keep a Changelog with BREAKING entries and migration notes for 3.0 and 4.0 and an UPGRADING-4.0.md, but 3.3.0, which relicensed the code, has no changelog section, and 4.0.0 is dated 19 September there while only betas exist (9 of 15). Stable is 3.5.1, after two earlier majors; 4.0.0 is in beta (15). 20 + 25 + 18 + 9 + 15 = 87.

Monid 53

Hosted lines, graded on the REST API and the remote MCP server it fronts, the surfaces an agent calls. status.monid.ai is a live Better Stack page, created 15 June 2026, with 90 days of daily history for the API (api.monid.ai), the dashboard and the landing page. It doesn't monitor mcp.monid.ai or x402.monid.ai. Snipcart earned 15 for a live page checking root URLs and Nevermined 10 for one that had stopped recording (15 of 20). From 26 July to 23 October 2026 (read at 12:15 UTC) the API was down 6,135 seconds, 1 hour 42 minutes, on 17 August, with no incident post, and the dashboard and site were clean. An hour or more of the core API down is one major outage (10). The OpenAPI file at monid.ai/openapi.json says limits apply per API key, user or workspace on every authenticated route and per provider and endpoint on runs, with numbers only in response headers. Paid earned 7 for limits described without numbers (7 of 15). 429s carry Retry-After in whole seconds and RateLimit-Limit, -Remaining and -Reset headers, and BLOCKED runs are documented as terminal. POST /v1/run spends money and takes no idempotency key, and nothing covers resending after a lost response. Paid earned 13 with an idempotency key on part of its writes and Nevermined 15 with one on payments (11 of 15). No SLA, and the terms promise no availability (0). No beta or preview label on the API or the MCP server (10).

MCP Gateway 90

4.0.0-beta.3 on 6 October 2026, 27 days before the check, and commits on main to 26 October (30). 4.0.0-beta.1 and beta.2 on 25 September and beta.3 on 6 October, plus 3.5.1 on 4 September, inside 90 days (20). The newest outside issue (#3953, 19 October) got a maintainer reply the same day and a follow-up on 28 October, four issues are open, and outside pull requests from four contributors were merged, per AUTHORSHIP.md. Closed issues and pull requests couldn't be read (robots.txt), so not sampled further (17 of 25). In the official MCP registry as io.github.MikkoParkkola/mcp-gateway under a GitHub namespace, latest 3.5.1, the current stable release (15). Dependabot merges, cargo audit in CI and current tokio and jsonschema, less 2 because the npm wrapper installs a downloaded binary without checking its checksum (8 of 10). 30 + 20 + 17 + 15 + 8 = 90.

Monid 64

Read for a hosted service with an open CLI and connector framework. @monid-ai/cli 0.1.8 reached npm on 21 October 2026, and the connector repository merged six OpenAI image endpoints the same day (30). CLI 0.1.6 on 27 July, 0.1.7 on 30 August and 0.1.8, and catalogue tags catalog-v0.0.1, catalog-v0.0.2 and catalog-v0.0.4 on 14, 17 and 26 September 2026 (20). The platform is closed and GitHub's web pages and API refused our reader, so issue queues and reply times are unread and the 0 to 15 scale for closed services applies, as in the Paid and Nevermined dossiers. No public changelog for the API, MCP server or CLI, support by Discord and email (not tested), and connector pull requests from outside authors merged, such as #27, #40, #50 and #64. Paid earned 5 with a support channel and no changelog, and Nevermined 8 with dated SDK changelogs, which Monid lacks (6 of 15). Not in the official MCP registry (a search for monid returned none) and no official SDK beyond the CLI (0). The connector repository runs typecheck and replay tests on pull requests and on main, keeps a lockfile and runs a weekly drift check against Apify's prices and schemas, and the CLI runs its tests before each npm publish. CI results weren't read (8 of 10).

MCP Gateway 80

Read with the API checklist, as the gateway is the security layer an operator puts in front of agents. Over HTTP, bearer tokens and API keys with per-key glob allow and deny lists, explicit admin keys, OIDC JWT and mTLS, with OAuth 2.1 to backends, per-issuer token storage and per-user isolation that fails closed. Authentication is off by default on the loopback HTTP port, where callers are anonymous non-admins, and a stdio caller is admin (24 of 30). Management tools need an admin credential since 3.5.0, per-key tool scopes give least privilege, and destructive meta-tools ask the human through elicitation, refused on the 2026-07-28 protocol when the client can't ask and allowed after a warning on older clients; the source calls the prompt a courtesy, with the admin requirement as the control (16 of 20). Backend tool descriptions are scanned for poisoning patterns before they reach the agent, capability files can be hash-pinned, and responses pass a pattern-based injection scanner and credential redactor (13 of 15). NDJSON audit log, per-key cost and usage accounting, Prometheus metrics and trace ids (13 of 15). SECURITY.md with private reporting, a 48-hour acknowledgement and a 7-day fix target; two advisories published with credit on 28 August 2026; cargo audit, secret scanning and cosign-signed images in CI; no bug bounty or security.txt, and the OWASP mapping is self-assessed (14 of 20). 24 + 16 + 13 + 13 + 14 = 80. The DNS-rebinding advisory is a negative event below.

Monid 62

API keys carry {resource}:{action} scopes over runs, resources, wallet, top-ups, controls and API keys, fixed at creation, from a published catalogue (GET /v1/api-keys/scopes). They can carry an expiry, are revoked in the dashboard or with POST /v1/api-keys/{label}/revoke, and a key can mint only keys within its own grants. Bearer header only, with no query-string option documented. OAuth 2.0 and OpenID Connect serve third-party apps and the MCP sign-in. That meets the line's scoped, revocable keys with rotation (30). Read-only keys are possible (runs:read, wallet:read), a person sets budgets per hour, day, week or month and a per-run cap, and keys minted from a dashboard session get no controls scope by default, so an agent holding one can't lift its own budget. There's no allowlist of providers or endpoints (the docs send builders to run their own proxy for that) and no approval step before paid runs or real-world actions such as calls and SMS. Nevermined and Composio each earned 16 (16 of 20). Runs return scraped pages, social posts and reviews from third parties, with no guidance on treating them as untrusted, and the skill tells agents to replace it from monid.ai on every update. Composio took 3 with none (3 of 15). Runs are kept with their caller, input, output, provider status, price and cost, readable through GET /v1/runs and per control, keys record who created and revoked them and when they were last used, resources have event histories, and every response carries an x-request-id. No retention period is stated (13 of 15). security.txt answers 404 on monid.ai, neither repository has a SECURITY.md, and no disclosure policy, bug bounty or certification was found. GitHub advisories couldn't be read, because GitHub's web pages and API refused our reader (0 of 20).

MCP Gateway 60

PolyForm Noncommercial 1.0.0, source-available and not OSI, with affirmative SPDX headers on every first-party file and LICENSES.md, NOTICE.md and COMMERCIAL.md that agree with each other and say what the earlier MIT grants still cover. Read as gaia's PolyForm source was (20 of 30). Self-hosted, so data stays where the owner runs it; secrets go to the OS keychain, files holding them are 0600 on Unix, and SECURITY.md states the Windows gap. There is no privacy statement (20 of 30). Deprecation notices with versions for the MIT 3.0.0 to 3.2.1 releases on crates.io and npm, upgrade guides for 3.0 and 4.0 and a stated support window of the latest release only, but the 3.3.0 relicence has no changelog entry, and AUTHORSHIP.md says the licensing scheme still awaits an IP lawyer's sign-off (9 of 20). No telemetry, analytics or update check found in the source; Prometheus metrics stay local behind a feature flag, and the plugin marketplace host is contacted only by plugin commands. The docs state nothing either way (15 of 20). 20 + 20 + 9 + 15 = 64.

Monid 47

The editorial half. The hosted platform, ranking, billing, API and MCP server are closed. The connector framework (definitions, compiler, engine and tests) is MIT on GitHub, and the CLI is MIT per its package.json and npm, with no LICENSE file in its repository. The home page's 'Monid is open source' sits under an invitation to contribute connectors, which the open repository bears out. The terms, last updated 16 October 2026, name Monid, Inc. but are a general template that forbids access 'through automated or non-human means' and purchases by 'a buying agent or purchasing agent', which the product is built for, and point to a Return Policy that wasn't found. Nevermined and Paid earned 17 for closed platforms with open SDKs under terms that didn't fit the product (17 of 30). The privacy policy, last updated 23 April 2026, covers monid.ai and app.monid.ai and says nothing about run inputs and outputs, their retention, the providers that receive them or training. It keeps data 'as long as necessary' or while an account exists, leaves its Cookie Notice link blank in three places, says no information comes from third parties while describing social sign-in, and permits third-party advertising trackers. The site loads Google Analytics, PostHog and an X pixel (5 of 30). The OpenAPI file marks POST /v1/discover deprecated in favour of /v1/discover/endpoints and resultCount as a deprecated alias, with no dates or policy, and the terms reserve the right to change the Services without notice. Paid earned 6 for six deprecated operations with named replacements and no dates (5 of 20). Stripe and Coinbase are named for payment data and the terms say the Services are hosted in the United States, but the analytics vendors aren't named, and the providers that receive each run's input are shown per endpoint in the catalogue, not as recipients in the policy (6 of 20). The provenance half is computed from the provenance block.

MCP Gateway 77

MCP context cost. Nine to 17 meta-tools, 11 in the default admin set, whatever the number of backends, so 15 for 11 to 30 tools, plus 10 back for Code Mode's two tools, exposed_meta_tools and surfaced tools (25). The project's own live run of 4 September 2026 found the meta path used 1.2 to 16.1 per cent more input tokens and one extra turn on every task, which the README states; the context saving is a schema-size saving, so less 5 (20 of 25). gateway_search_tools and gateway_search take limit, gateway_search has L0, L1 and L2 detail tiers, gateway_list_tools filters by server and role, and backend responses can be shaped with transform chains (17 of 20). Did-you-mean on tool names, LLM-readable output-schema failures, circuit-breaker and disabled status in gateway_list_servers, and rate-limited calls excluded from the kill budgets (16 of 20). Meta-tools carry readOnlyHint, destructiveHint, idempotentHint and openWorldHint; destructive meta-tools ask through elicitation, which the source calls a courtesy rather than a control; an idempotency guard exists per ADR-012 (16 of 20). One required argument for search and two for invoke; Rust only, no SDK applies, and setup takes a wizard and a config file (8 of 15). 20 + 17 + 16 + 16 + 8 = 77.

Monid 71

Read on the REST API, since the MCP tool list is unread. Discover returns 5 results by default, up to 200, above a score floor and narrows by category, so the catalogue stays out of context. Run outputs come back whole, sized only through each provider's own parameters, so 5 off (20 of 25). List routes page by cursor and filter by status, category, provider, x402 eligibility and resource effects, while run output size depends on each provider's limits, which the skill warns often apply per query rather than per call (18 of 20). Errors carry code, message and an optional errorCode. A 402 from Monid is always the caller's wallet, with a provider's own 402 shown as 502, run statuses separate Monid failures, control blocks and provider statuses, and responses carry hints naming the next command. No retryable flag, which Nevermined's 20 rested on (17 of 20). No idempotency key on runs, and the MCP annotations are unread. Endpoints carry readonly and effects metadata on resource changes, and resource release is documented as idempotent. Smithery earned 10 with idempotent upserts and safe client retries, and Monid's endpoint metadata covers less than annotations (8 of 20). A run needs only provider, endpoint and input, with sensible defaults, but there's no official SDK in any language, only the CLI and raw HTTP examples (8 of 15).