"Finding MCP servers that support SSE or Streamable HTTP, excluding stdio transport" matching MCP tools:

• install-matrixA

  MeshMarketC

  Install Matrix — Generate copy-paste-correct MCP install snippets for ~29 clients at once — with each client's config-key traps already encoded (Antigravity demands `serverUrl` and lowercase names; Gemini CLI demands `httpUrl`; AnythingLLM demands type 'streamable'; Goose/Kiro/Cursor/VS Code get working deeplinks; ChatGPT gets the search+fetch requirement spelled out). Use when you or your user needs to wire ANY MCP server into a client without hunting per-client docs. Deterministic, no model call. Input: {server_url: string (required, http(s) URL), name?: string, transport?: 'streamable-http'|'sse', auth?: 'none'|'bearer-optional'|'bearer-required'}. Returns {clients: [{client, method: 'config-file'|'cli'|'deeplink'|'paste-url', snippet, config_path?, notes?}], count}. (1 MESH/call, a tool · devtools)

  ConnectorNo auth

• registry_totalsA

  agent-economy-dataA

  Returns how many MCP servers the Official MCP Registry lists, split by status (active, deprecated), remote transport (streamable-http, sse) and package type (npm, pypi, ...). Example output: {"unique_servers":53618,"by_status":{"active":52778,"deprecated":840}}. No arguments. Good for 'How many MCP servers are there, and how many run remotely?'.

  ConnectorNo auth

• check_mcp_healthA

  CodeSentinelA

  Probe a remote MCP server: Streamable HTTP / SSE handshake (initialize + tools/list), a known-bad tools/call error-shape probe, and a Streamable HTTP diagnostic matrix (WRONG_METHOD, WRONG_ACCEPT, MISSING_SESSION, GET_VS_POST, SESSION_STICKY_MISMATCH). HTTP 200 is not treated as healthy.

  ConnectorAPI key

• find_agentsA

  AgentXMB

  Ranked agents for a task. Constraints: auth (any|none|oauth), transport (any|streamable-http|sse), max_latency_ms, max_cost_per_success_usd, min_works. Pass stack as a comma-separated list of tools you already use (e.g. 'github, postgres') to get a Fit boost, or a profile_key from the site. Keep the query_id and call report_outcome after using a result.

  ConnectorNo auth

• guild_checkA

  agent-guildB

  START HERE when asking "which agent should I hire for this capability?" One call to vet a `capability` before you delegate: returns the best-evidenced agent with an evidence verdict — `estimate` (0-1), `confidence`, and a checkable `explanation` — plus a ranked shortlist, machine-checkable PROOF the Guild improves outcomes (provenance-labelled), and how to contribute back. Read estimate AND confidence together and apply your own risk threshold: a high estimate with low confidence means thin evidence. This is a PAID trust read (same price + policy as GET /check on every transport). When the rail is active, an unpaid call returns a complete x402 payment challenge for the canonical HTTP resource; retry with the payment in the request _meta['x402/payment'] (official x402 MCP meta key). Schema-driven agents that cannot set request metadata may pass that same signed PaymentPayload as `x402_payment`. Or pass a funded `api_key` for SANDBOX credits (never revenue). Free while the service is in soft-launch. Example: guild_check(capability="fact-check") Returns {capability, best_agent, verdict, shortlist, proof, why_trust_this, how_to_contribute}. Use guild_search / guild_risk_score for finer control.

  ConnectorNo auth

• list_investment_thesesA

  crashtestyourstrategyA

  Discover the investment-thesis catalog. Each entry is a descriptive case study that pairs an economic framework with a rule-based portfolio and the synthetic + historical stress evidence for that allocation. Returns one compact summary per thesis (slug, title, one-liner, tags, risk tiers, framework summary, headline finding). Call get_investment_thesis(slug) for the full framework / portfolio / stress evidence, or read the thesis://{slug} resource. Descriptive, not advisory — the agent decides what is suitable.

  ConnectorNo auth

• upload_design_fileA

  a2a2p — Agent-to-Agent-to-PhysicalA

  Turn local manufacturing file bytes into the temporary HTTPS reference used by specification.design_files. This is a bounded intake operation, not a request, provider call, quote, order, payment, or fabrication action. Common RFC 4648 transport forms are normalized, then decoded bytes are recognized by signature, checked against the declared format, screened before storage, retained for one hour, and never included in telemetry. Supported roles include geometry (STL/STEP) and drawings (DXF/PDF). The 2 MiB MCP ceiling is transport-specific; use POST /api/upload with purpose=manufacturing_design_file for files up to 8 MiB. Pass the returned file_reference object unchanged in specification.design_files on the next call.

  ConnectorNo auth

• account_support_accessA

  AMZ VaultB

  Grant or revoke the AMZ Vault support team's READ-ONLY view of this account: action=grant lets support see what the user sees for 7 days (auto-expires, they can change nothing, and every view notifies the user); action=revoke turns it off immediately. This changes who can access the account, so it ALWAYS needs the user's explicit go-ahead — offer it only when support needs to see a problem first-hand (usually alongside an account_feedback thread).

  Connector

  Destructive

  OAuth

• explain_findingA

  MalinoisA

  Returns the plain-language meaning and step-by-step fix for one Malinois finding. Use it while helping the user fix an issue reported by scan_app, or when they ask what a finding means. Pass the rule_id exactly as scan_app returned it. Read-only, no network, instant.

  ConnectorNo auth

• save_payment_linkA

  financeA

  Save (or update) the external payment URL for one payment context. Use when the user gives or corrects a link — e.g. "for child support use https://childsupport.state.mn.us". Saving a label that already exists (case-insensitive) UPDATES that link's URL. Intended for URLs the user explicitly provided or confirmed, not found, guessed, or constructed ones. https only; a bare domain like "chase.com" is normalized to https://; javascript:/http:/IP-literal/credential-bearing URLs are rejected. Args: label: The payment context name (max 100 chars), e.g. "child support". url: The user-provided payment page URL (https). Returns: {"success": True, "created": bool, "payment_link": {...}} — created is False when an existing label's URL was updated.

  ConnectorAPI key

• record_finding_verdictA

  aictrlA

  Apply a post-hoc triage verdict to a finding that was previously recorded via record_finding. Used by the /reply-to-code-review flow to persist whether a finding is TRUE / FALSE / UNCERTAIN and the FIX / DEFER / IGNORE action taken. The finding is located only by the exact `finding_id` returned by list_pr_findings, on ANY lifecycle status (open, fixed, or dismissed — so verdicts still apply after a PR is merged and findings were swept to dismissed). Only findings recorded into the data layer (the aictrl-dev MCP path) can receive a verdict. Idempotent: re-applying overwrites the prior verdict. A not_found response is an application error; call list_pr_findings to refresh stale IDs and do not treat the write as successful. `judged_at_sha` is required: send the full 40-hex commit SHA of the code you actually read. A SHA that is neither the review SHA nor a later commit of the same PR is rejected with validation_error. `reasoning` is required too (#5364): a verdict whose reasoning cannot be audited later is not worth storing. Each write appends to the finding's judgment history; a re-judge from the same run at the same commit updates that run's own row rather than appending a duplicate. A FALSE verdict on an OPEN finding of severity MINOR or NIT (medium/low/info) also dismisses it as a false positive, with your `reason` as the note; on MAJOR or BLOCKER (high/critical) it stays a suggestion a person confirms. The result reports `autoDismissed` (and `autoDismissReason` when false). A dismissal is not reopened by this tool; a person can restore the finding from the pull-request page.

  Connector

  Destructive

  API key

• panda_get_ai_ebook_upload_urlA

  Panda VideoB

  Get a temporary upload URL (valid 1 hour) to send a file of the eBook: kind=pdf — your own PDF as the eBook of a language that has no eBook yet (up to 15 MB); kind=logo — a PNG/JPEG logo for the eBook (up to 5 MB). The URL is a credential: don't show or share it. Send the file with an HTTP PUT to upload_url with the Content-Type header exactly as returned. Then: PDF → panda_confirm_ai_ebook_pdf; logo → pass logo_key in ebook_config of panda_update_ai_resource (existing eBook) or panda_generate_ai_resource. Your client must be able to send files over HTTP; otherwise use the dashboard (Video Settings → Panda AI → eBook).

  ConnectorNo auth

• update-variables-toolA

  YouformA

  Replace a form's custom variables — named values that logic rules read and write (scores, prices, outcome tallies). Pass the COMPLETE custom list from list-logic-options-tool, excluding entries with built_in true. Names use letters, numbers and underscores only; type is "number" (starts at 0) or "text" (starts empty). Built-in and field names are reserved. Create every variable BEFORE referencing it in set-block-logic-tool. Changes apply to the live form immediately. To remove or change the type of a variable in use, remove its references from the draft and publish first.

  Connector

  Destructive

  OAuth

• stream_taskA

  TunnelMind Data APIB

  Opens a persistent SSE connection that emits events as the task progresses. The stream closes automatically when the task reaches a terminal state or after ~90 seconds (timeout). Heartbeat comments are sent every ~15 seconds to keep the connection alive through proxies. Event types: - `status` — emitted when status changes (pending → running → complete/failed) - `result` — emitted on `complete` with the full result payload - `error` — emitted on `failed`, `cancelled`, or `expired` with error info - SSE comment (`: heartbeat`) — keepalive, no data Use this tool when: - You want real-time progress without polling. - You are in an environment that supports SSE (EventSource API). Do NOT use this tool when: - You want a simple one-shot status check — use `get_task` instead. - Your HTTP client doesn't support streaming responses. Inputs: - `task_id` (path, required): 26-char ULID. Returns: - SSE stream (`text/event-stream`). Each event is `event: <type>\\ndata: <json>\\n\\n`. Cost: - Free. Counts as one request against rate limits when the stream opens. Latency: - First event: <200ms. Stream duration: up to 90s.

  ConnectorNo auth

• link_issueA

  DeskCrew - AI Chatbot, Helpdesk & Knowledge baseA

  Associate an existing support ticket with a tracked issue (bug or feature) in the same tenant, so the ticket is connected to the underlying work item. Reach for this when a customer's ticket is caused by, or asks for, a known issue and you want to record that relationship. This records an issue_link entry on the ticket's timeline (optionally with a note) but does NOT change the ticket's status or notify the customer; both the ticket and the issue must already exist in your tenant or the call fails with NOT_FOUND. [price: $0.03]

  ConnectorNo auth

• check_endpointA

  cerberus-indexB

  Endpoint & tool trust — one call before an agent connects to a third-party MCP server or HTTP tool. Pass the endpoint `url` (and, to unlock the strongest check, its `tools`: the name/description/inputSchema the agent is about to trust). Returns three signals behind one verdict: (1) transport & TLS identity — encrypted, valid chain, not expired or self-signed; (2) domain age via RDAP — freshly-registered hosts are a scam tell; (3) a TOOL-POISONING scan of the tool definitions for the hidden directives that hijack agents — instruction overrides, 'don't tell the user', data exfiltration, secret harvesting, tool-shadowing, and invisible-unicode / homoglyph steganography that a human reviewer can't see. Nobody else screens tool descriptions for injection. Verdict: trusted | caution | untrusted | unknown, with per-finding evidence. Price: $0.008 USDC. Missing something? Call `submit_feedback` (free) to request it. Pass `attest=true` to also get an Ed25519-SIGNED attestation of this verdict — portable proof you can log, hand to a counterparty, or verify later with `verify_attestation` (free) or the published key.

  ConnectorNo auth

• show_tours_activitiesA

  GuruWalkA

  Present a selection of tours or activities to the traveler as a visual list of cards (photo, rating, price, booking button) rendered inline in the conversation. Showing cards is the default way to present tours to the traveler: whenever your reply features specific tours (recommendations, a shortlist, availability results), call this tool alongside your text instead of waiting to be asked. Don't re-render a selection you already showed unless it changed. Call it AFTER finding tours with the search tools — it is a presentation tool, not a search tool. If you already checked availability with get_product_availability, you can pass each item's sessions (date and start time) and the cards will highlight them with booking links preselecting the date. For clients without UI support the same data is returned as structured text.

  ConnectorNo auth

• rooms_postA

  grith-mcpA

  Opens a room (send title) or speaks in one (send id and body). Rooms are group chat, kept permanently. For the public board use plaza_post; for one person use message. Body up to 800 characters. 20 an hour; a burst up to three times that is parked, beyond that refused. Proof: send your citizen secret as HTTP Authorization: Bearer, or in the leave_token argument if your host cannot set headers, or a controller-key signature over a fresh nonce (nonce, controller_public_key, controller_signature). Writes go to grithgate.com; grithland.com is look-only and refuses them.

  ConnectorNo auth

• rooms_postA

  GRITH MCPA

  Opens a room (send title) or speaks in one (send id and body). Rooms are group chat, kept permanently. For the public board use plaza_post; for one person use message. Body up to 800 characters. 20 an hour; a burst up to three times that is parked, beyond that refused. Proof: send your citizen secret as HTTP Authorization: Bearer, or in the leave_token argument if your host cannot set headers, or a controller-key signature over a fresh nonce (nonce, controller_public_key, controller_signature). Writes go to grithgate.com; grithland.com is look-only and refuses them.

  ConnectorNo auth

• get_facility_staff_rosterA

  Georgia Prisoners' Speak Public DataA

  Staff headcount and list for one GDC facility, from a GDC-produced roster. Answers "how many staff work at X", "how many officers work there", "list the staff at X". `slug` is the facility slug, e.g. 'dooly-state-prison' (use list_facilities to find it). `title` optionally filters by job title. GDC abbreviates heavily: correctional officers appear as "Correctional Ofc", clerical staff as "Admin Support" or "Admin Assistant". Try both the plain word and the abbreviation. ALWAYS report the `observed` date alongside any figure from this tool. A roster is a single-day observation, not a statement about who works there today. A zero for a role is a finding about the roster rather than missing data: the roster holds no nursing, physician or dental titles anywhere in GDC, which is consistent with medical staffing being contracted rather than directly employed. Do not assert that reason as established fact.

  ConnectorNo auth

• encode_urlA

  Razi ToolsA

  Percent-encode a string with encodeURIComponent so reserved characters survive transport inside a URL. Returns JSON { encoded }. It escapes the structural characters too — : / ? # & = all become %XX — so it is for a single query-string value or path segment, NOT for a whole address you still want to be clickable. This is escaping for URLs only. It is the wrong transform for making binary-unsafe content fit in JSON, an HTTP header or a data URI, which call for Base64 rather than percent-encoding. Reverse it with decode_url.

  ConnectorNo auth