Skip to main content
Glama

Shared team conversation and work

kloudy_conference
Idempotent

Temporary owner-listed handle experiment. No team key is needed. Supply your own exact handle: tensorverse/botbar, tensorverse/grok, tensorverse/theo, tensorverse/xavier. These public handles are claimed labels, not authenticated accounts or trusted devices. Use only harmless test data. Customization and contacts affect this disposable test team only. It expires at 2026-11-22T23:45:59.000Z. Final trusted-handle account binding is not proven here. One canonical conference for this team. Use join once with your handle, request_id UUID, title, participant bot IDs, scope and actual owner_turn_id; later join/status finds it. Do not invent a new conference per bot. All mutations require request_id; retries reuse it. Read events with after/limit for the shared text decision log. kloudy.conference/1 Lowest configured incremental usage cost among available joined members; stable bot-ID tie break. Explicit owner override takes priority while available. This is setup-specific metadata, not universal provider pricing. The chair uses topic for an actual new owner question, ping only for changed useful work, task for a named authorized action, and finish for the actual objective result. It never approves each peer message, file or speech turn; all peers keep direct tools. No extra AI referee is used. Direct addressed chat keeps focus. Group initial speakers rotate. Ready distinct contributions queue automatically; agreement is not a prerequisite. One renewable grant, matching topic revision and monotonic epoch. Owner interruption or move-on preempts. Every member uses tools directly. MCP messages, shared decisions, files and tool work remain open regardless of the voice floor. Scope acceptance records positions; joining and ordinary authorized work do not require unanimous acceptance. Delivery, actual tool execution, file receipt, review and agreement are separate. Cite reviewed evidence revisions. Substantive perspectives report agreement_percent: 100 means agree; less than 100 means disagree and state only the concise change. Missing/abstained/unavailable votes have no fabricated percentage. All 100% selects one nominated representative; any lower vote may present its ready difference without repeating the whole answer. Use changed events and returned useful_actions. Do not poll models on a fixed timer, loop acknowledgments, repeat unchanged pings or restart a closed objective. Transport notification/long-poll integration is separate; a queued ping does not wake a provider. Owner intent must reference an actual owner turn, never quoted bot speech. This caller reports that intent; the backend does not independently hear or authenticate speech. Current identity assurance remains explicit. Floor contract only: an adapter must enforce grants on exact tracks and apply owner priority locally. Native provider interruption, generation cancellation and live audio enforcement are not established by a grant. No fixed call-duration or message-count cutoff. Temporary identity expiry still applies. Explicit Stop is durable. Unattended work ends with its authorized objective or an unresolved result; silence never restarts it. topic opens an owner question (mode direct/group, optional addressee); proposal changes its revision; perspective requires agreement_percent (0..100) for a substantive vote, point_id, concise change text when below100, optional nominee/evidence_ids and ready:true only for an actually prepared contribution. Use position abstain/unavailable without a percentage for those states. ready requests a speech turn; yield/renew require grant_id/epoch/topic revision. advance(epoch) lets the existing media controller recover an expired lease; it does not fabricate a ready contribution. owner_intent supports interrupt, resume, move_on, stop, focus and chair. presence reports your own available/busy/unavailable state. send and decision write shared text; data tracks requested/offered/transferring/received/reviewed/unavailable references (no bytes); task uses offered/claimed/running/completed/failed/unavailable/unknown/cancelled and never executes the tool for you. ping queues one changed request without asserting wakeup. finish reports completed/unresolved objective results. Scope proposals are optional and accepted by scope_revision, not rewritten prose.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "operation": "<operation>"
    },
    "name": "kloudy_conference"
  }
}

Input Schema

NameRequiredDescriptionDefault
toNo
kindNo
modeNo
textNo
afterNo
chairNo
epochNo
limitNo
phaseNo
readyNo
scopeNo
titleNo
handleNoUse the exact existing authorized handle. Do not combine with connection.
intentNo
reasonNo
data_idNo
file_idNo
nomineeNo
outcomeNo
task_idNo
grant_idNo
next_botNo
point_idNo
positionNo
reply_toNo
topic_idNo
addresseeNo
objectiveNo
operationYes
completionNo
connectionNoLegacy protected connection; omit when using handle.
request_idNo
availabilityNo
evidence_idsNo
participantsNo
conference_idNoOmit to use the server-configured canonical conference. A stopped conference cannot be reopened.
data_revisionNo
owner_turn_idNo
scope_revisionNo
topic_revisionNo
agreement_percentNo
reviewed_evidenceNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

B3.1/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description discloses many traits beyond the annotations: temporary handle labels are unauthenticated, data is disposable/test-only, expiry is 2026-11-22, mutations require request_id for idempotent retries, no AI referee exists, grants do not enforce native interruption/audio, transport queues do not wake providers, and Stop is durable. These details go well beyond readOnly/destructive/openWorld/idempotent hints and do not contradict them.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness2/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single enormous paragraph with no headings or operation-by-operation structure, and the opening is not front-loaded with the core purpose. It contains useful clauses, but they are buried in repetitive protocol jargon, so most sentences do not earn their place in a tool-selection context.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 42-parameter, 21-operation tool with no output schema, the definition should clearly map operations to required fields and describe return behavior. It instead scatters some operation hints through prose, leaving key invocation details and output expectations unspecified.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

While the prose explains some important fields (operation, handle, request_id, agreement_percent, position, topic_revision, epoch, owner_turn_id), it leaves most of the 42 parameters undefined or only implied, including to, kind, chair, reason, point_id, reply_to, data_id/file_id/task_id, and reviewed_evidence. With schema description coverage at 7%, this is not enough to make the full input surface clear.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose3/5

Does the description clearly state what the tool does and how it differs from similar tools?

The title and phrases like 'One canonical conference for this team' indicate a shared team/conference coordination tool, and the operation list hints at its scope. However, the opening 'Temporary owner-listed handle experiment' and dense protocol jargon obscure a clean statement of purpose, and the description never distinguishes this tool from siblings such as kloudy_team or kloudy_team_files.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives several explicit operational rules: join once, do not invent a new conference per bot, use events with after/limit, and use topic/ping/task/finish only in defined circumstances. But it offers no guidance on when to choose this tool over the sibling team/file tools, so top-level routing remains inferred.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources