MCP Verification Gate: check an MCP server or an A2A agent before you connect or delegate
REMOTE · GATE.HORIZONSHIELD.DEV
Check an A2A agent before you delegate, or an MCP server's measured conduct. Free, no key.
Available components
Last scanned (13 hours ago)
Scores higher than 92% of remote servers
Reachable. Last reached 24 Oct 2026.
Category Security
How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. How we score → Why this is hard to score →
Endpoint Security71
- The endpoint's TLS certificate is valid, in date, and uses a strong key. View diagnostics → Pass
- No authorisation is required to call this server. Every tool declares its destructiveHint and none is destructive, so open access doesn't expose one. See how to fix → View diagnostics → Partial
- HTTPS enforcement could not be verified: the plaintext port answered with HTTP 405, which proves neither a plaintext path nor enforcement. View diagnostics → Unverified
- The HSTS (Strict-Transport-Security) header is present. View diagnostics → Pass
- DNSSEC is configured correctly; the domain's records validate against the full chain to the root. View diagnostics → Pass
Transport & Reachability100
- Verified streamable-http transport via a live MCP handshake. View diagnostics → Pass
Schema Quality & AI Usability79
- AI-judged instruction clarity (excellent).Pass
- Context-footprint check failed: tool/resource definitions use about 815 tokens (~135/item across 6 items; 6 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management100
- No destabilizing schema changes in the last 30 days.Pass
Tool Coverage100
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 100% of tool parameters carry a description.Pass
- Structured output schemas are declared (100% of tools); any adoption earns full credit.Pass
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- We read all 6 captured tool definition(s), and no name or description among them implies an irreversible operation.Pass
- An AI judge read all 6 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.Pass
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 10 Oct 26 0
- HTTPS: fail → unverified ▼ security
- Tool coverage: 83% → 100% ▲ functional
- 28 Sept 26 +3
- DNSSEC: fail → pass ▲ security
- HSTS header: fail → pass ▲ security
- Tool coverage: 100% → 83% ▼ functional
- Schema quality: pass → fail ▼ functional
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
- Server version: 0.4.18 → 0.4.20 functional
- Server version: 0.4.17 → 0.4.18 functional
- 26 Sept 26 0
- Server version: 0.4.14 → 0.4.15 functional
- 25 Sept 26 0
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
- 21 Sept 26 82
- Server version: 0.4.7 → 0.4.14 functional
- 11 Sept 26 0
- Server version: 0.4.4 → 0.4.7 functional
- 10 Sept 26 0
- Server version: 0.4.2 → 0.4.4 functional
- 9 Sept 26 0
- Server version: 0.4.0 → 0.4.2 functional
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 24 Oct 2026 · Probed https://gate.horizonshield.dev/mcp
TLS valid
Negotiated TLS 1.3 with TLS_AES_128_GCM_SHA256 .
| Subject | Issuer | Valid from | Valid until | Key | Signature | Serial |
|---|---|---|---|---|---|---|
| CN=horizonshield.dev | CN=WE1,O=Google Trust Services,C=US | 9 Oct 2026 | 7 Jan 2027 | ECDSA 256 | ECDSA-SHA256 | ec78a88f90aee500133fc3a9a556a9b1 |
| SANs: horizonshield.dev, *.horizonshield.dev | ||||||
| CN=WE1,O=Google Trust Services,C=US (CA) | CN=GTS Root R4,O=Google Trust Services LLC,C=US | 13 Dec 2023 | 20 Feb 2029 | ECDSA 256 | ECDSA-SHA384 | 7ff31977972c224a76155d13b6d685e3 |
| CN=GTS Root R4,O=Google Trust Services LLC,C=US (CA) | CN=GlobalSign Root CA,OU=Root CA,O=GlobalSign nv-sa,C=BE | 15 Nov 2023 | 28 Jan 2028 | ECDSA 384 | SHA256-RSA | 7fe530bf331343bedd821610493d8a1b |
Background: What to check on a remote MCP endpoint →
DNSSEC secure
Validation of gate.horizonshield.dev. — Secure
| Zone | DS | Keys | Algorithms | Outcome |
|---|---|---|---|---|
| . | trust_anchor | 20326, 38696 | 8, 8 | Verified |
| dev. | present | 60074 | 8 | Verified |
| horizonshield.dev. | present | 2371 | 13 | Verified |
| gate.horizonshield.dev. | Verified address RRset verified with the apex keys |
Authentication No authorisation required
The endpoint answered without asking for a token. Anyone who knows the URL can reach it.
| Result | No authorisation required |
|---|---|
| HTTP status | 200 |
| Header | Value |
|---|---|
| strict-transport-security | max-age=31536000; includeSubDomains; preload |
| x-content-type-options | nosniff |
Background: How OAuth 2.1 works in the 2026 MCP spec →
Transports 2 probes
| Transport | URL | Outcome | Status | Location |
|---|---|---|---|---|
| streamable-http | https://gate.horizonshield.dev/mcp | Verified | 200 | |
| http (plaintext) | http://gate.horizonshield.dev/mcp | Inconclusive | 405 |
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
check_conformance Check an MCP server for conformance and disclosure ~186
Measure a public MCP endpoint against five conditions: it speaks MCP, it publishes an A2A agent card, it declares who pays it, identical input returns identical output, and the verdict itself can be recomputed by anyone. Free, no key. Conformance and disclosure only; this says nothing about whether any figure the checked server returns is correct. By default no tool on the checked server is called, so determinism comes back as not measured rather than guessed. It is measured when the owner has published /.well-known/mcp-conduct.json with allow_tool_call true.
| Name | Type | Req | Description |
|---|---|---|---|
| allow_tool_call | boolean | – | Kept for compatibility. Since 2026-10-08 an assertion alone calls no tool: consent is proven only by the owner's /.well-known/mcp-conduct.json. Default false. |
| endpoint | string | yes | https URL of the MCP endpoint to measure |
| Name | Type | Req | Description |
|---|---|---|---|
| absence_vs_failure | object | – | Condition 06, disclosed and never a verdict: whether the server's tools can tell a failed lookup from an empty one. |
| canonicalization | object | – | Condition 07, disclosed and never a verdict: whether the declared tool surface canonicalizes under RFC 8785. |
| checked_at | string | – | When the measurement started, ISO 8601 UTC. |
| checks | object | – | One entry per condition, each with whether it was measured, whether it passed, and the detail. |
| consent_assertion_ignored | object | – | Present when the request set allow_tool_call true without proven consent: the assertion was recorded and no tool was called. |
| consent_basis | string | – | On what basis a tool call was or was not made. |
| consent_lookup | object | – | Present when no proven consent was found: the consent file that was read, the result, and how to consent. |
| consent_source | string | – | Where the consent came from: operator_list, well_known, requester or none. |
| coordinate_derivation | object | – | How the measured tool and instant were derived, or that the legacy rule applied. |
| does_not_establish | array | – | What this verdict does not establish, as sentences. |
| endpoint | string | – | The MCP endpoint that was measured. |
| establishes | array | – | What this verdict establishes, as sentences. |
| gate | string | – | Name of this gate. |
| gate_commit | string | – | Git commit of the deployed gate, or a sentence saying the deployment did not pin one. |
| gate_version | string | – | Version of the gate code that took this measurement. |
| measurement_note | string | – | Present only when the gate's own relay failed, so that nothing here is read as a statement about the target. |
| number_safety | object | – | Whether every number in this verdict survives a JSON round trip unchanged. |
| probed_via | string | – | The network path the probe took (relay or direct). |
| reachable | boolean|null | – | Three-valued on purpose (gate58). true = measured and answered. false = measured and did not answer. null = NOT MEASURED. null is never to be read as a failing endpoint; it means this gate has nothin… |
| recompute_note | string | – | How to recompute record_sha256. Not part of the hashed bytes. |
| record_sha256 | string | – | Hash of this verdict with record_sha256 and recompute_note removed. Recompute it yourself; verify_verdict does the same arithmetic. |
| scope_note | string | – | What this measurement covers and what it does not. |
| status | string | – | verified = every measured condition passed. pending = measured, not all passing. held = could not be reached, nothing established. |
| tools_called | string | – | Whether a tool on the checked server was executed, in words. |
No examples provided.
get_conditions Get the conformance conditions ~56
Return the five conditions this gate measures, what it explicitly does not verify, and the tier definitions. Takes no arguments and returns identical output every time. Read this before running a check so you know what a verdict does and does not claim.
Input schema present but exposes no named parameters.
| Name | Type | Req | Description |
|---|---|---|---|
| also_measured_no_verdict | object | – | Measurements that are disclosed and never change a verdict. |
| conditions | object | – | The measured conditions, keyed by condition id, each with what is measured and how. |
| consent | string | – | When this gate calls a tool on a checked server, and what counts as the owner's consent. |
| establishes_and_does_not_establish | object | – | What a verdict establishes and what it does not, as two lists. |
| gate | string | – | Name of this gate. |
| gate_commit | string | – | Git commit of the deployed gate, or a sentence saying the deployment did not pin one. |
| instant_coordinate | object | – | How the time of a scheduled measurement and the measured tool are derived rather than chosen. |
| lookup | object | – | How to read the register for one endpoint before connecting (GET /register/lookup). |
| number_safety | object | – | How numbers in a verdict are kept recomputable across JSON implementations. |
| operator | string | – | Who operates this gate. |
| reachability | string | – | How an answer from the server is told apart from a failure to reach it (pending versus held). |
| record_bytes | object | – | Where the exact bytes that record_sha256 hashes are served, and since when. |
| red_team | string | – | The adversarial test suite this gate is run against, with its scores by version. |
| self_applied | string | – | Statement that this gate is measured under its own conditions. |
| tiers | object | – | The status words a verdict can carry and what each one means. |
| version | string | – | Version of the gate code that produced this document. |
| well_known_consent | object | – | The consent file an origin can publish, its path and its fields. |
| what_this_does_not_verify | array | – | What a verdict from this gate never supports, as sentences. |
| what_this_verifies | array | – | The claims a verdict from this gate supports, as sentences. |
No examples provided.
is_verified One glance: is this MCP server verified, with proof ~147
A single machine-first answer for an agent deciding whether to trust an MCP endpoint BEFORE it connects. Returns verified (true only when the latest scheduled measurement passed every measured condition; null otherwise, never false), a state enum saying which case it is, measured_at, and a record_sha256 with a recompute_url so you can check the verdict without trusting this gate. Reads the stored register only: it contacts nothing and measures nothing. Absent and pending are reported honestly and are NOT negative verdicts. For a fresh measurement rather than the stored one, use check_conformance.
| Name | Type | Req | Description |
|---|---|---|---|
| endpoint | string | yes | https URL of the MCP endpoint to look up, exactly as it appears on the register |
| Name | Type | Req | Description |
|---|---|---|---|
| absence_vs_failure | object|null | – | Measured endpoints only: condition 06 from the latest measurement, disclosed and never a verdict. |
| conditions | object|null | – | Per condition: true passed, false did not pass, null not measured. null as a whole when nothing was measured. |
| endpoint | string | – | The endpoint you asked about, echoed. |
| gate | string | – | Name of this gate. |
| gate_commit | string | – | Git commit of the deployed gate, or a sentence saying the deployment did not pin one. |
| history_url | string | – | URL of the stored measurements for this endpoint. |
| how_to_appear | string|null | – | Absent endpoint only: how anyone can add it to the register. |
| measured_at | string|null | – | Time of the latest stored measurement, or null when there is none. |
| measurements | integer | – | Measured endpoints only: how many measurements are stored. |
| not_an_endorsement | string | – | What a verified reading does not claim. |
| on_register | boolean | – | true = the endpoint has a row here. false = the register was read and it has none. |
| reason | string | – | One sentence saying why state has the value it has. |
| recompute_note | string | – | How to recompute record_sha256 from record_url, or why it cannot be recomputed. |
| recompute_url | string | – | Where to fetch what record_sha256 can be checked against: record_url when it exists, otherwise the history. |
| record_sha256 | string|null | – | Hash of the latest verdict. Recompute it via recompute_url; no trust in this gate required. |
| record_url | string|null | – | URL of the exact bytes that record_sha256 hashes, or null for a verdict stored before those bytes were kept. |
| state | string | – | verified = passed all measured conditions. pending = measured but not passing every one (often only because determinism needs the owner's consent). held = could not be reached. watched = on the list,… |
| verified | boolean|null | – | true = latest measurement passed all measured conditions. null = not established here (see state). Never false: unmeasured or not-yet-passing is not a failure. |
| verified_meaning | string | – | What verified true and null mean. |
No examples provided.
lookup_server Look up an MCP server on this register ~145
Returns one record (not a list). Look up what this register already holds about an MCP endpoint: whether it is watched, how often it is re-measured, how many measurements exist, when the first and latest were taken, and the latest verdict with the record_sha256 you can recompute yourself. Reads stored measurements only. It contacts nothing and measures nothing, so use check_conformance for a fresh reading. An endpoint that is absent is reported as absent and that is NOT a negative verdict: it means nobody has measured it here, not that it failed.
| Name | Type | Req | Description |
|---|---|---|---|
| endpoint | string | yes | https URL of the MCP endpoint to look up, exactly as it appears on the register |
| Name | Type | Req | Description |
|---|---|---|---|
| added_at | string|null | – | On the register: when the row was added, or null if not recorded. |
| alerted_on_change | boolean | – | On the register: whether someone asked to be notified when the row changes. |
| cadence | string | – | On the register: how often it is re-measured (weekly or daily). |
| does_not_mean | string | – | What this answer must not be read as. |
| endpoint | string | – | The endpoint you asked about, echoed. |
| first_measured_at | string|null | – | On the register: time of the first stored measurement, or null. |
| fresh_reading | string | – | Absent endpoint only: how to measure it right now. |
| full_history_url | string | – | On the register: URL of the complete history for this endpoint. |
| history | array | – | On the register: the most recent stored measurements, oldest first, at most 20. |
| history_truncated | boolean | – | On the register: true when history leaves out older measurements. |
| how_to_appear | string | – | Absent endpoint only: how anyone can add it to the register. |
| last_measured_at | string|null | – | On the register: time of the latest stored measurement, or null. |
| latest | object|null | – | On the register: the latest stored measurement as a summary with its record_sha256, or null if none. |
| means | string | – | What this answer means. |
| measurements | integer | – | On the register: how many measurements are stored for this endpoint. |
| on_register | boolean | – | true = this endpoint is on the register. false = the register was READ and this endpoint is not on it. If the register could not be read at all, this field is not returned: the call comes back as a t… |
| register_size | integer | – | Absent endpoint only: how many endpoints are on the watchlist. |
| standing | string | – | On the register: measured, or a sentence saying it is watched and not yet measured. |
| tier | string | – | On the register: free, paid or self. The verdict is the same for every tier. |
No examples provided.
preflight_agent Before delegating: check an A2A agent's card, who pays it, and its register reading ~187
Call this before handing work to an A2A agent you have not used before. Fetches the agent's public card (https://<agent>/.well-known/agent-card.json, one request), reports whether it declares the A2A Conduct Extension, who it says pays it (the compensation declaration, as declared and not verified), whether the card carries a signature, the endpoints it asks to be measured on, and this register's stored reading for each (verified is true only when the latest scheduled measurement passed; null otherwise, never false), plus where to file your own witness walk. Measures nothing new and returns counts and pointers, never a score. An agent that does not declare the extension is reported as such, which is not a negative verdict.
| Name | Type | Req | Description |
|---|---|---|---|
| agent | string | yes | https origin of the agent (https://agent.example) or the full URL of its agent card |
| Name | Type | Req | Description |
|---|---|---|---|
| agent | string | – | The agent's https origin. |
| card_signature_present | boolean | – | Whether the card carries a signature. Presence only; the signature is not verified here. |
| card_status | integer|null | – | HTTP status the card URL answered with. |
| card_url | string | – | The agent card URL that was fetched. |
| compensation | object|null | – | Who the agent says pays it, as declared and not verified, or null when it says nothing. |
| compensation_source | string|null | – | Where the compensation declaration was read: extension params or card top-level. |
| conduct_record | string|null | – | Where the card says its conduct record is published, or null. |
| declared_uri | string|null | – | The extension URI the card declared, or null. |
| does_not_establish | array | – | What this reading does not establish, as sentences. |
| extension_declared | boolean | – | true = the card declares the A2A Conduct Extension. false is not a negative verdict. |
| how_to_file | string | – | How to walk the agent yourself and file your own record. |
| measured_endpoints | array | – | The https endpoints the card asks to be measured on, at most 5. |
| ok | boolean | – | Always true in this object: the card was fetched and read. |
| register | array | – | This register's stored reading for each of those endpoints: state, verified (true or null, never false), record_sha256. |
| schema | string | – | Format name of this object: gate-preflight-v1. |
| witness_intake | string|null | – | Where the card says witness records can be filed, or null. |
No examples provided.
verify_verdict Recompute a verdict hash without trusting the issuer ~94
Take a verdict this gate issued and recompute its record_sha256 independently. Removes record_sha256 and recompute_note, serialises the remainder in key order, and hashes it. Returns whether the verdict was altered after it was issued. You do not have to trust the party that issued the verdict, including this one.
| Name | Type | Req | Description |
|---|---|---|---|
| record | object | yes | The full verdict object as returned by check_conformance or GET /self |
| Name | Type | Req | Description |
|---|---|---|---|
| expected_sha256 | string|number|boolean|object|array | – | The record_sha256 found in the record you passed, echoed as given (a string in any verdict this gate issued). Absent when the record had none. |
| method | string | – | The arithmetic used, so you can repeat it without this gate. |
| note | string | – | What a match or a mismatch does and does not prove. |
| reason | string | – | Present instead of the hashes when the input could not be checked: not an object, or no record_sha256. |
| recomputed_sha256 | string | – | SHA-256 this call computed from the record. Absent when there was nothing to compare it with. |
| verified | boolean | – | true = the verdict hashes to its own record_sha256, so it was not altered after issue. false = it was altered, or it carries no record_sha256. This is a finding about the record, not an error. |
No examples provided.
How do I install the MCP Verification Gate: check an MCP server or an A2A agent… server?
MCP Verification Gate: check an MCP server or an A2A agent… is a hosted endpoint at https://gate.horizonshield.dev/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
remote · gate.horizonshield.dev
claude mcp add --transport http ogasurfproject-jpg-hs-verify-gate 'https://gate.horizonshield.dev/mcp'
What is the MCP Verification Gate: check an MCP server or an A2A agent… server?
MCP Verification Gate: check an MCP server or an A2A agent… is listed in the public MCP registry as io.github.ogasurfproject-jpg/hs-verify-gate. Check an A2A agent before you delegate, or an MCP server's measured conduct. Free, no key. This page covers its hosted endpoint (https://gate.horizonshield.dev/mcp).
Is the MCP Verification Gate: check an MCP server or an A2A agent… server safe to use?
MCP Verification Gate: check an MCP server or an A2A agent… scores 85 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the MCP Verification Gate: check an MCP server or an A2A agent… server expose?
MCP Verification Gate: check an MCP server or an A2A agent… exposes 6 tools: get_conditions, check_conformance, verify_verdict, lookup_server, is_verified, preflight_agent. Their descriptions and schemas cost roughly 815 tokens of context every time the server is loaded.
Does the MCP Verification Gate: check an MCP server or an A2A agent… server require authentication?
No. We connected to MCP Verification Gate: check an MCP server or an A2A agent… without credentials and it answered, so anything it exposes is reachable by anyone who knows the address.
Is the MCP Verification Gate: check an MCP server or an A2A agent… server still maintained?
MCP Verification Gate: check an MCP server or an A2A agent… is still listed as active in the MCP registry. We last reached this channel on 24 October 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.