Skip to main content
Glama
kaelorlabs

Smitline

by kaelorlabs
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Start a phone call or join a meeting

start_call

Initiate an AI voice call or video meeting to fulfill an objective, returning a structured result with answers and decisions when done.

Instructions

Place a phone call or join a video meeting for the user. Smitline talks with people in real time using GPT-Live and returns a structured result when the call ends. Write a complete brief: the goal, what to find out (questions), what may be agreed to, and what must not be shared. Pass what you and the user have been working on as context (summary, facts, and long details) so the assistant can answer questions like someone who knows the story. The user's profile (who they are, the people they know) is added automatically; keep it current with update_profile. If anything required is unknown, ask the user instead of guessing. For a first call to someone new, offer a rehearsal on the user's own phone. Returns immediately; then call wait_for_call.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
toNoE.164 phone number such as +14155550142, or the meeting invite URL. Omit only for a rehearsal, which rings the user's own phone.
taskNoThe job several calls serve, such as getting quotes from ten roofers. Use the same id for each call in it.
toneNoHow to come across, such as "casual, he is a close friend". Defaults to matching the relationship.
voiceNoGPT-Live voice name; see list_voices
notifyNo
recordNoPhone only: record this call in the user's phone account; the assistant tells the other person the call is recorded. Only when the user asks for a recording.
channelYesphone to place a call; meeting to join Zoom, Teams, or Google Meet
contactNoWho is being called, when the user's profile does not already know this number.
contextNoWhat you and the user have been working on that the other party may ask about. Text, or an object: summary (a few sentences), facts, decisions, openQuestions, and details (long reference material). The voice starts with a short version and looks details up when asked; it never recites them.
languageNoLanguage tag such as en or es
carryFromNoStart with short notes from earlier calls (at most 10). task: earlier calls in this brief's task; contact: earlier calls to the same number (phone only), or false to skip what the user switched on for that contact; calls: these call ids. Only carry what this call needs: the assistant may mention it to the other party, so do not carry one business's details to another unless the objective calls for it.
objectiveYesWhat the call must achieve, in one or two sentences
questionsNoWhat to find out on the call. The result answers each one.
rehearsalNoPhone only: practice on the user's own phone first, with the user playing the other side
afterHoursNoPhone only: call even though it is outside calling hours (8 AM to 9 PM) where the person is. Set only when the user confirms the person expects a call now.
maxMinutesNoPhone calls: at most 60 (default 10). Meetings: at most 240 (default 120).
mayAgreeToNoWhat the assistant may agree to without checking back, such as acceptable times or prices
onBehalfOfNoThe user's name, spoken in the opening: Hi, I'm calling on behalf of NAME about... Defaults to the name given at setup.
mustNotShareNoInformation the assistant must never share
successCriteriaNoHow to tell the call succeeded

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.5/5.0
Behavior5/5

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

Annotations cover safety (readOnlyHint=false, destructiveHint=false, openWorldHint=true), and the description adds substantial context beyond them: the call is asynchronous ('returns immediately'), a structured result arrives at call end, the user's profile is injected automatically, recording must be disclosed to the other party, and calling hours are 8 AM–9 PM. This is unusually rich disclosure for a mutation tool.

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

Conciseness4/5

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

Purpose is front-loaded and the operational guidance (brief contents, next tool call) is packed into a compact block, though the 'Smitline talks with people in real time using GPT-Live' sentence is flavor rather than instruction. Slightly longer than ideal for an 8-sentence description but every remaining sentence carries actionable weight.

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

Completeness4/5

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

For a 20-parameter, nested-object, no-output-schema tool, the description covers the async lifecycle, the brief contract, profile auto-injection, and hand-off to wait_for_call. It does not describe what the eventual structured result contains, but that responsibility plausibly sits with wait_for_call/get_call, so the gap is minor.

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

Parameters4/5

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

Schema coverage is already 95%, so the baseline is 3. The description goes further by prescribing what a complete brief must contain (goal, questions, may agree to, must not share) and explaining the purpose of context, which maps directly to several parameters. It adds real meaning rather than restating field descriptions.

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

Purpose5/5

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

States a specific verb and resource ('Place a phone call or join a video meeting') and covers both channels named in the enum. An agent can distinguish this from siblings like wait_for_call and transfer_call_to_me without opening the schema.

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

Usage Guidelines4/5

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

Explicitly routes the agent to the next step ('Returns immediately; then call wait_for_call'), points to update_profile for profile upkeep, list_voices for voice selection, and describes the rehearsal path for first-time contacts. It lacks an explicit 'when not to use' exclusion versus adjacent tools like send_call_instruction, so it stops short of 5.

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

Deploy Server

Other Tools