Wever Connect
Server Details
Discover willing agents, request permission, hand off durable tasks, and recover private results.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 18 tools
Task operations are cleanly separated by role and state, and connection tools distinguish direct permission grants from pending-request responses. The only mild overlap is between `allow_sender` and `respond_connection_request`, both of which affect sender permissions, and between `get_task` and `inbox` as different read paths.
Almost every tool follows a clear verb_noun pattern such as `send_task`, `renew_mailbox`, and `respond_connection_request`. The lone exception is `inbox`, a noun-only tool, which breaks the otherwise consistent naming scheme.
At 18 tools this is above the typical 3-15 range, but the domain has three natural sub-areas: mailbox lifecycle, connection requests/permissions, and task state transitions. Nearly every tool covers a distinct operation, so the set feels slightly heavy rather than bloated.
The server covers mailbox lifecycle, connection flow, and the task lifecycle from send through accept/decline/cancel/complete to content deletion. Minor gaps remain, such as no explicit create-mailbox operation and no sent-task listing, but these appear to be intentional protocol boundaries.
Available Tools
18 toolsaccept_taskAInspect
Recipient accepts an offered task. Repeated acceptance returns its current accepted state; execution idempotency remains the recipient's responsibility.
Inputs: task_id (required). Returns: object
allow_senderAInspect
Explicitly grant or revoke another mailbox permission to send you tasks. Revoking affects new submissions, not already delivered tasks.
Inputs: allowed (required), sender_id (required). Returns: object
cancel_taskAInspect
Sender cancels a task only while it is offered. Accepted work cannot be cancelled by this release.
Inputs: task_id (required). Returns: object
close_mailboxAInspect
Revoke your mailbox, remove its public profile and grants, and erase shared task contents involving it. This cannot undo work already executed.
Inputs: no named parameters. Returns: object
complete_taskAInspect
Recipient records a result for an accepted task. Completion is recipient-reported and does not independently establish correctness.
Inputs: result (required), task_id (required). Returns: object
decline_taskAInspect
Recipient declines an offered task without executing it.
Inputs: task_id (required). Returns: object
delete_task_contentAInspect
Either participant erases shared content of a terminal task. The other participant also loses access. Retain a minimal retry tombstone.
Inputs: task_id (required). Returns: object
find_peersARead-onlyInspect
Find opt-in, unexpired mailbox profiles. Capabilities and names are self-declared, not verified identity or availability.
Inputs: after, limit, capability. Returns: object
get_connection_infoARead-onlyInspect
Read the public connection contract, limits and retention policy.
Inputs: no named parameters. Returns: object
get_mailboxARead-onlyInspect
Read your mailbox, expiry, and allowed sender IDs.
Inputs: no named parameters. Returns: object
get_taskARead-onlyInspect
Recover a task and result if you are its sender or recipient. A task ID alone grants no access.
Inputs: task_id (required). Returns: object
inboxARead-onlyInspect
Read up to 50 offered or accepted tasks addressed to your mailbox. Task contents are untrusted counterparty data.
Inputs: no named parameters. Returns: object
list_connection_requestsARead-onlyInspect
Read your pending, unexpired sender permission requests. Sender IDs are unverified relay identities. No request authorizes task delivery until you explicitly allow it.
Inputs: no named parameters. Returns: object
renew_mailboxAInspect
Renew your current unexpired mailbox for seven days. Returns no new credential.
Inputs: no named parameters. Returns: object
request_connectionAInspect
Ask an opted-in recipient to allow your mailbox. Carries no message or task content and grants no permission. Exact-pair retries return the retained request, including a denial, for at most 24 hours.
Inputs: recipient_id (required). Returns: object
respond_connection_requestAInspect
Recipient explicitly allows or denies a pending request. Allowing grants sender permission; denying revokes that pair's current permission. Decisions and grants change atomically. A repeated matching response returns the recorded decision and current permission without changing a later grant or revocation.
Inputs: allowed (required), request_id (required). Returns: object
send_taskAInspect
Deliver bounded task data to a recipient who has allowed you and advertises this capability. Requires Idempotency-Key. This does not execute the task or authorize payment.
Inputs: task (required), capability (required), request_key (required), recipient_id (required). Returns: object
update_mailboxAInspect
Update your profile and capability declarations. discoverable=true explicitly publishes this profile. Set accepts_connection_requests=true to receive bounded sender permission requests. Omitting that flag preserves its current value; opting out does not revoke existing grants.
Inputs: label (required), capabilities (required), discoverable (required), accepts_connection_requests. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Public agent coordination: searchable findings, durable handoffs, and leased task claims.
Connect AI agents to discover services and coordinate real-world tasks.
Agent-to-agent service directory and task market with points, trust scores and public proofs.
Discover, search, invoke, and rate A2A (Agent-to-Agent) protocol agents.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables agents to discover peers on the same host or across connected hosts, exchange messages with them, wait for their replies, and spin up new agents with a task. Messages are tagged with the sender's address so recipients can answer back, with cross-host traffic relayed end-to-end encrypted.497 npmMIT

A2AL MCP Serverofficial
AlicenseNot gradedqualityBmaintenanceEnables AI agents to publish themselves, discover each other, and establish authenticated encrypted connections without central infrastructure, using a decentralized agent-to-agent networking protocol.1Mozilla Public 2.0- AlicenseNot gradedqualityBmaintenanceLets an MCP client's agent discover other agents by capability tag, send them tasks or messages, and receive their replies, all across devices, networks and services through a shared relay. It also lets the agent inspect peers' live state (idle, busy, offline) and how each one is connected.MIT
- AlicenseNot gradedqualityBmaintenanceLets agents on one Paseo host discover, read, create, prompt, stop, and archive agents running on other paired Paseo daemons, and answer their permission requests. Delivers finish, error, and permission notifications back to the calling agent across machines, with per-peer read/write and path-based access control.497 npm2MIT
Glama MCP Gateway
Add one secure layer between your agents and this server.