Skip to content

Agentic Endpoints MCP server

com.agentic-endpoints/agentic-endpointsagentic-endpoints.comstreamable-httpv1.0.0Scanned

Pay-per-call tools for autonomous agents, settled in USDC on Base via x402.

24 tools: 3 can delete or cancel things, 1 do something that can't be undone, 12 change things and 8 only read.

Tools: 3 delete, 1 irreversible, 12 write, 8 read.
  • Reads private data · 6
  • Reads text others write · 3
  • Can send data out · 5
  • 3 findings

Tools listed without sign-in. Remote posture

With Anlyon

3 tools here can delete or cancel things and 1 does something that can't be undone. Govern them.

  • Anlyon lists the tools and adopts none.
  • Each tool you adopt gets approvals and receipts.
  • Your agent never holds the key.
Govern this server →

A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.

What the Agentic Endpoints MCP server can do

The Agentic Endpoints MCP server is published by agentic-endpoints.com, a domain the MCP Registry verified. It has 24 tools. It can delete or cancel things (3 tools: vault_store, vault_delete and vault_rotate_token), do things that cannot be undone, such as sending, paying or publishing (reach) and create or change data (12 tools: x402_verify, meetings_import, once_key_claim, once_key_complete and 8 more). 8 tools only read.

6 tools read private data (meetings_search, meetings_summarize, meetings_get and 3 more); 3 tools bring in text other people can write (pdf_parse, scrape and reach); 5 tools can send data out of your system (x402_verify, pdf_parse, scrape and 2 more). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.

The fixed checks matched 3 findings (annotation-conflict and tool-shadowing). Each one comes with the text it matched.

Its remote endpoint lists its tools without sign-in.

Last scanned Nov 12, 2026, version 1.0.0.

Tools

Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.

24 tools

Deletes or cancels 3Claude Code: deny
  • vault_deletePermanently delete an item.No findings

    Why it is in this group

    • The name "vault_delete" says "delete": a word for removing or ending something.
    • The description says "Permanently delete".
    • The server marks it destructive (destructiveHint: true).
    Description, as the server wrote it
    Permanently delete an item. Requires the namespace_token. Costs $0.005 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: true
    • idempotentHint: true
    • openWorldHint: false
  • vault_rotate_tokenFree.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it destructive (destructiveHint: true).
    Description, as the server wrote it
    Free. Mints a new namespace_token and immediately invalidates the current one, which you must present to authorize the rotation. Do this whenever the token may have been exposed — a leaked token is otherwise permanent, unrevocable read and delete access to every secret in the namespace. There is no recovery if you lose the token. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: true
    • idempotentHint: false
    • openWorldHint: false
  • vault_storeStore client-side encrypted data.No findings

    Why it is in this group

    • The name "vault_store" says "store": a word for a change.
    • The description says "cannot be recovered".
    • The server marks it destructive (destructiveHint: true).
    Description, as the server wrote it
    Store client-side encrypted data. This service holds no key that could decrypt it and never sees plaintext; note that the item key, namespace, alg label and size ARE stored in the clear. The first store claims the namespace and returns a namespace_token shown only once — store it immediately, because it is required by every later call and cannot be recovered. Pass if_match with an item's updated_at for a compare-and-swap write, or if_absent to create only; either returns status 'precondition_failed' rather than silently clobbering a concurrent write. Costs $0.02 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: true
    • idempotentHint: true
    • openWorldHint: false
Can't be undone 1Claude Code: ask
  • reachSend one GET to a URL as each of nine common clients (curl, node, Python urllib, Python requests, libwww-perl, a Chrome browser, GPTBot, ClaudeBot, Googlebot) and report which are …Reads text others writeCan send data out1 finding

    Why it is in this group

    • The description's first verb is "send": a word for sending, and what is sent cannot be taken back.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe description's first verb is "send": it sends or publishes something to other people.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Warn

      The server marks "reach" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

      Matched text

      annotations: {"readOnlyHint":true,"idempotentHint":true,"openWorldHint":true}; The description's first verb is "send": a word for sending, and what is sent cannot be taken back.

      What this check looks for
    Description, as the server wrote it
    Send one GET to a URL as each of nine common clients (curl, node, Python urllib, Python requests, libwww-perl, a Chrome browser, GPTBot, ClaudeBot, Googlebot) and report which are refused and why, such as a Cloudflare 1010 block on Python's urllib. Also reports robots.txt blocks on AI crawlers, whether /llms.txt exists, and whether the http-to-https redirect keeps POST bodies. Use before relying on a site or API from code, or to check your own. Read-only; private addresses are refused. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: true
Changes things 12Claude Code: ask
  • compressReduce text to fit a token budget, preserving whole sentences and reporting before/after token estimates.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Reduce text to fit a token budget, preserving whole sentences and reporting before/after token estimates. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • credits_balanceFree.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Free. Reports the remaining balance on a credit token so you can tell 'out of credit' apart from 'the call failed', which otherwise look identical from the outside. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • credits_trialFree, and the first tool to call if any other tool has told you payment is required.1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.

    Findings

    • Warn

      The description contains "any other tool", wording about other tools or servers.

      Matched text

      description: "Free, and the first tool to call if any other tool has told you payment is required. Retur..."

      What this check looks for
    Description, as the server wrote it
    Free, and the first tool to call if any other tool has told you payment is required. Returns a credit_token carrying $0.10 of credit — roughly ten calls at the standard price — with no wallet, no signature, no account and no email. Pass the returned token as the `credit_token` argument to any paid tool and it will run without an x402 payment. One allowance per caller: asking again returns the same token and whatever balance is left on it, never a refill. When it runs out, buy more with the /credits/buy endpoint. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • meetings_importStore a transcript so it can be searched later.No findings

    Why it is in this group

    • The name "meetings_import" says "import": a word for a change.
    Description, as the server wrote it
    Store a transcript so it can be searched later. Accepts a raw WebVTT or SRT export straight from Zoom, Teams, Meet or a notetaker -- you do NOT need to strip timestamps or parse it first, and cues split across timing boundaries are merged so phrases still match. `visibility` is a required decision and cannot be guessed for you: 'queryable' stores plaintext, indexes it, and means this service can read it; 'private' stores ciphertext you encrypted yourself, which is unreadable here and therefore NEVER searchable. Send `transcript` for queryable and `ciphertext` for private -- the mismatched combinations are refused rather than silently doing the wrong thing. The first import into a namespace returns a namespace_token shown exactly once; save it or the namespace is unrecoverable. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • once_key_claimAtomic idempotency witness.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Atomic idempotency witness. Claims a {namespace, action_key} pair exactly once, so a fleet of agents cannot perform the same side effect twice. Call this BEFORE any non-idempotent action such as sending an email, charging a card, or posting an order. Returns one of: 'claimed' — you won, do the work, then call once_key_complete; 'in_progress' — another agent holds a live lease, wait retry_after seconds and do NOT do the work; 'duplicate' — already done, and the 'result' field carries the original outcome, so use it instead of repeating the work; 'held' — another agent claimed this key, set no lease, and has NOT completed it: there is no result and there may never be one, so do NOT do the work and do NOT treat it as done, because the key stays locked until expires_at; 'conflict' — the same key was claimed with a different payload hash, so your key derivation is wrong. Backed by a strongly consistent Durable Object; this is not something an agent can safely reimplement locally. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • once_key_completeFree.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Free. Records the result of work you performed under a claim from once_key_claim. Every later claim of that action_key returns 'duplicate' along with this result, which is what lets another agent continue without repeating the side effect. Always call this after the work succeeds — a claim with no recorded result leaves every other agent unable to learn what happened. Completion is final and cannot be overwritten. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • once_key_releaseFree.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Free. Surrenders a claimed action_key so a retry can start immediately instead of waiting out the lease. Call this when the work you claimed fails. Refuses if the key was already completed, because releasing it would discard the recorded result and allow the side effect to run twice. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • pdf_parseExtract text from a PDF by URL.Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    Extract text from a PDF by URL. Handles compressed streams, PDF 1.5+ object streams and ToUnicode CMaps, and reports encrypted or image-only documents honestly instead of returning garbage. Costs $0.020 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: true
  • reach_reportOrder a full reachability report for one site: up to 20 pages, each checked from nine common clients.Can send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.

    What it can reach

    • outboundThe input takes a URL (input.urls): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    Order a full reachability report for one site: up to 20 pages, each checked from nine common clients. Returns a report_id and report_url at once; the pages are scanned in the background, about ten seconds each. The finished report gives the likely cause of every refusal with exact fix steps and a reproducing curl command, a prioritised fix list, and three free re-scans within 30 days. Use reach_report_status to read it. For a single quick check use the reach tool instead. Costs $69.00 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • reach_report_rescanFree, up to three times within 30 days of ordering.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Free, up to three times within 30 days of ordering. Re-runs every page of a finished report so you can confirm a fix worked; the report then shows what was fixed, what remains and anything new. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • vault_existsCheck for a key without returning its ciphertext.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Check for a key without returning its ciphertext. Requires the namespace_token. Costs $0.005 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • x402_verifyCall this BEFORE authorising payment to any x402 endpoint you did not write yourself.Can send data out1 finding

    Why it is in this group

    • The description's first verb is "call": a word for a change.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Warn

      The server marks "x402_verify" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

      Matched text

      annotations: {"readOnlyHint":true,"idempotentHint":true,"openWorldHint":false}; The description's first verb is "call": a word for a change.

      What this check looks for
    Description, as the server wrote it
    Call this BEFORE authorising payment to any x402 endpoint you did not write yourself. It fetches the endpoint's live payment challenge and compares it against every observation previously made by every other caller, so you learn things a single agent cannot see on its own -- above all, whether the address receiving the money has changed. A 'critical' entry in `drift` means something determining where funds go (pay_to, network or asset) is different from before: stop and confirm out of band. Read the LIVE `charges` in preference to any directory listing, and pass what the listing claimed as `expect` to have the disagreement reported. IMPORTANT: 'status':'ok' is NOT an endorsement -- it only means a challenge was read and recorded. It cannot tell you whether the operator will deliver anything for your money. 'first_observation':true means there is no history at all, so an empty `drift` proves nothing. ALWAYS read `prior_criticals`: `drift` only compares against the last observation, so an endpoint that swapped its payee earlier shows an empty `drift` once that swap became the baseline -- a non-zero `prior_criticals` means it has changed where money goes before, and that is true no matter how clean today's look is. 'status':'unreachable' is not evidence of fraud and not evidence of health; do not pay on it either way. A null `price_usd` means the token's units are unknown here and the amount was NOT converted -- do not assume it is small. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
Only reads 8Show 8 read-only tools
  • meetings_getFetch a single meeting by meeting_id, including its full transcript when it was imported as queryable.Reads private dataNo findings

    Why it is in this group

    • The name "meetings_get" says "get": a word for reading.

    What it can reach

    • private-dataThe name "meetings_get" says "get" and "meeting": it reads calendars.
    • private-dataThe description's first sentence says "fetch" and "meeting": it reads calendars.
    Description, as the server wrote it
    Fetch a single meeting by meeting_id, including its full transcript when it was imported as queryable. Use this after meetings_search has identified the meeting you want. A 'content_missing' status means the record exists but its stored text could not be found -- that is a broken record, not an empty meeting, so do not report it as one. Costs $0.005 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • meetings_listList the meetings in a namespace newest first, with titles, dates, participants and whether each one is searchable.Reads private dataNo findings

    Why it is in this group

    • The name "meetings_list" says "list": a word for reading.

    What it can reach

    • private-dataThe name "meetings_list" says "list" and "meeting": it reads calendars.
    • private-dataThe description's first sentence says "list" and "meeting": it reads calendars.
    Description, as the server wrote it
    List the meetings in a namespace newest first, with titles, dates, participants and whether each one is searchable. Never returns transcripts. Useful for orienting before a search, and for finding meetings that are private and therefore invisible to meetings_search. Costs $0.005 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • meetings_summarizeAsk a plain-language question and get a written answer grounded in the user's own meeting transcripts, with the meeting id cited after each claim.Reads private dataNo findings

    Why it is in this group

    • The name "meetings_summarize" says "summarize": a word for reading.

    What it can reach

    • private-dataThe name "meetings_summarize" says "summarize" and "meeting": it reads calendars.
    • private-dataThe description's first sentence says "meeting": it reads calendars.
    Description, as the server wrote it
    Ask a plain-language question and get a written answer grounded in the user's own meeting transcripts, with the meeting id cited after each claim. Prefer this over meetings_search when the user wants an ANSWER ('what did we decide about pricing?', 'who owned the migration?') rather than a list of excerpts to read themselves. Pass the question as the user asked it -- this is NOT FTS5 syntax, and the terms actually searched come back as `terms` so you can check the retrieval. IMPORTANT: a 'no_matches' status means no answer was generated, and 'searched_meetings': 0 means every meeting is private and NOTHING was read -- neither is evidence the topic was never discussed, and you must not report it as such. Meetings imported as private are encrypted and can never contribute to an answer; `private_meetings_skipped` says how many were left out. A 'truncated' meeting in `consulted` had only part of its transcript read, so its silence on a point is not evidence either. Costs $0.050 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • reach_report_statusFree.No findings

    Why it is in this group

    • The name "reach_report_status" says "status": a word for reading.
    Description, as the server wrote it
    Free. Returns a reachability report by its report_id: progress while it scans, then the prioritised findings with causes, fix steps and reproducing commands, the per-page evidence, and a comparison with the previous scan after a re-scan. This tool is free; no payment is required.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • scrapeFetch a URL and return readable text or markdown, optionally narrowed by CSS selector.Reads text others writeCan send data outNo findings

    Why it is in this group

    • The description's first verb is "fetch": a word for reading.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • untrusted-inputThe name "scrape" says "scrape": it reads web pages, which anyone can write.
    • untrusted-inputThe description's first sentence says "fetch" and "url": it reads web pages, which anyone can write.
    • untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Claude Code: ask, because it is tagged outbound.

    Description, as the server wrote it
    Fetch a URL and return readable text or markdown, optionally narrowed by CSS selector. Requests to private, loopback and link-local addresses are refused, and every redirect hop is re-validated. Costs $0.010 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: true
  • vault_listList the keys held in a namespace with their metadata: alg, size, timestamps.Reads private dataNo findings

    Why it is in this group

    • The name "vault_list" says "list": a word for reading.

    What it can reach

    • private-dataThe name "vault_list" says "list" and "vault": it reads secrets.
    Description, as the server wrote it
    List the keys held in a namespace with their metadata: alg, size, timestamps. Never returns ciphertext — use vault_retrieve for that. Each item's updated_at is the version to pass back as if_match on a conditional store. Use this when you have stored secrets and need to know what is there. Costs $0.005 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false
  • vault_retrieveRetrieve a previously stored ciphertext.Reads private dataNo findings

    Why it is in this group

    • The name "vault_retrieve" says "retrieve": a word for reading.

    What it can reach

    • private-dataThe name "vault_retrieve" says "retrieve" and "vault": it reads secrets.
    Description, as the server wrote it
    Retrieve a previously stored ciphertext. Requires the namespace_token. Costs $0.02 in USDC on Base, paid via the x402 protocol, or from a credit token — call credits_trial for free credit if you have neither.

    Annotations the server declares

    • readOnlyHint: true
    • idempotentHint: true
    • openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

90 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 3 untrusted-input tools breaks every one.

Reads private data6
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write3
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out5
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
deny to break every path
{
  "permissions": {
    "deny": [
      "mcp__agentic-endpoints__pdf_parse",
      "mcp__agentic-endpoints__reach",
      "mcp__agentic-endpoints__scrape"
    ]
  }
}

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Remote posture

What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.

Tools listed without sign-in

Looked at Nov 20, 2026, 18:33 UTC, unchanged since Oct 22, 2026. No credentials were sent and no tool was called. What is looked at.

Reachable
yes
HTTP status of an anonymous initialize
200
tools/list served without credentials
yes
Tools that change things, listed without credentials
16
Protocol version
2025-06-18
WWW-Authenticate on 401
not observed
resource_metadata in WWW-Authenticate
not observed
RFC 9728 protected resource metadata
not observed
Authorization servers
not observed
Legacy HTTP+SSE transport
no
Dynamic client registration
not observed
Client ID metadata documents
not observed
  • HighThe whole server

    The endpoint served tools/list without credentials, and the list has 16 tools that change things (3 destructive, 1 irreversible, 12 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://agentic-endpoints.com/mcp: tools/list returned 24 tools to an anonymous client, including vault_store (destructive), vault_delete (destructive), vault_rotate_token (destructive), reach (irreversible), x402_verify (write) and 11 more

    What this check looks for

Findings

Matches of fixed checks, with the text they matched.

3 warn
  • The server marks "x402_verify" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

    Matched text

    annotations: {"readOnlyHint":true,"idempotentHint":true,"openWorldHint":false}; The description's first verb is "call": a word for a change.

    What this check looks for
  • Warnreach

    The server marks "reach" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

    Matched text

    annotations: {"readOnlyHint":true,"idempotentHint":true,"openWorldHint":true}; The description's first verb is "send": a word for sending, and what is sent cannot be taken back.

    What this check looks for
  • The description contains "any other tool", wording about other tools or servers.

    Matched text

    description: "Free, and the first tool to call if any other tool has told you payment is required. Retur..."

    What this check looks for

History

Changes to its tool definitions and to its remote endpoint, newest first.

No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.

Install

Pinned to version 1.0.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http agentic-endpoints https://agentic-endpoints.com/mcp
claude mcp add --transport http agentic-endpoints https://agentic-endpoints.com/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__agentic-endpoints__meetings_search",
      "mcp__agentic-endpoints__meetings_summarize",
      "mcp__agentic-endpoints__meetings_get",
      "mcp__agentic-endpoints__meetings_list",
      "mcp__agentic-endpoints__vault_retrieve",
      "mcp__agentic-endpoints__vault_list",
      "mcp__agentic-endpoints__reach_report_status"
    ],
    "ask": [
      "mcp__agentic-endpoints__x402_verify",
      "mcp__agentic-endpoints__meetings_import",
      "mcp__agentic-endpoints__once_key_claim",
      "mcp__agentic-endpoints__once_key_complete",
      "mcp__agentic-endpoints__once_key_release",
      "mcp__agentic-endpoints__vault_exists",
      "mcp__agentic-endpoints__pdf_parse",
      "mcp__agentic-endpoints__scrape",
      "mcp__agentic-endpoints__reach",
      "mcp__agentic-endpoints__reach_report",
      "mcp__agentic-endpoints__reach_report_rescan",
      "mcp__agentic-endpoints__compress",
      "mcp__agentic-endpoints__credits_trial",
      "mcp__agentic-endpoints__credits_balance"
    ],
    "deny": [
      "mcp__agentic-endpoints__vault_store",
      "mcp__agentic-endpoints__vault_delete",
      "mcp__agentic-endpoints__vault_rotate_token"
    ]
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as agentic-endpoints. Review it before you use it.

Capability tags behind 11 rules
  • mcp__agentic-endpoints__meetings_search private-data
  • mcp__agentic-endpoints__x402_verify outbound
  • mcp__agentic-endpoints__meetings_summarize private-data
  • mcp__agentic-endpoints__meetings_get private-data
  • mcp__agentic-endpoints__meetings_list private-data
  • mcp__agentic-endpoints__vault_retrieve private-data
  • mcp__agentic-endpoints__vault_list private-data
  • mcp__agentic-endpoints__pdf_parse untrusted-input, outbound
  • mcp__agentic-endpoints__scrape untrusted-input, outbound
  • mcp__agentic-endpoints__reach untrusted-input, outbound
  • mcp__agentic-endpoints__reach_report outbound

Pin

Keep the decision: fail CI when the tools change, follow every change, or show a badge.

Pin in CI

Fail a build when the tools you reviewed are not the tools the server serves today.

curl -s "https://radar.anlyon.com/api/v1/servers/com.agentic-endpoints/agentic-endpoints/verify?hash=a17f2cffc063ca3648fdf369bd8309ffaf204b8762e88ac6cfba625e94a79a02" | jq -e '.changed == false'

Pinned hash a17f2cffc063, from Nov 12, 2026, 12:18 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/com.agentic-endpoints/agentic-endpoints/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for com.agentic-endpoints/agentic-endpoints
[![Anlyon Radar](https://radar.anlyon.com/badge/com.agentic-endpoints/agentic-endpoints.svg)](https://radar.anlyon.com/servers/com.agentic-endpoints/agentic-endpoints)

Source: MCP Registry. Synced Oct 21, 2026, 17:48 UTC.

Agentic Endpoints MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the Agentic Endpoints MCP server safe?

No scan can promise that. What its definitions show: 24 tools, 16 of which can change, send or delete something, and 3 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the Agentic Endpoints MCP server delete data?

Yes. 3 tools can delete, cancel or revoke something: vault_store, vault_delete and vault_rotate_token.

Can the Agentic Endpoints MCP server do things that cannot be undone?

Yes. 1 tool can send, pay, publish or merge: reach.

What tools does the Agentic Endpoints MCP server have?

24 tools: 3 destructive, 1 irreversible, 12 write, 8 read. vault_store, vault_delete, vault_rotate_token, reach, x402_verify, meetings_import, once_key_claim, once_key_complete and 16 more.

Does the Agentic Endpoints MCP server need sign-in or OAuth?

Its remote endpoint listed its tools to an anonymous client, without sign-in.

Does the Agentic Endpoints MCP server send data to outside services?

5 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): x402_verify, pdf_parse, scrape, reach and reach_report.

Has the Agentic Endpoints MCP server changed its tools?

Not since Anlyon Radar started tracking it. Last scanned Nov 12, 2026.