TestMyVibes

Run a two-AI voice conversation through Paradise Comms

submit_conversation_test

Spawns two AI personas as participants on a real LiveKit voice call (via Paradise's self-hosted comms cluster), driven by PM Brain's realtime lane by default, and runs a structured conversation. Each persona talks aloud (TTS) and listens to the other (Whisper STT) — this isn't simulation, it's a real WebRTC call with real audio. Used to verify Paradise Comms end-to-end (publisher → SFU → subscriber → recording → outbound webhook) and to demo agent-to-agent voice. Returns the full transcript and a recording hint.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`turns`	No	Total back-and-forth turns. 6 means A→B→A→B→A→B. Cap of 20 to bound LLM + TTS + Whisper spend per test.
`personaA`	No	Persona for agent A (speaks first). Defaults to skeptical-cto. See lib/personalities.ts for the full list of 8 personas.	skeptical-cto
`personaB`	No	Persona for agent B. Defaults to power-user. The pairing skeptical-cto + power-user is the canonical demo because their voices contrast strongly enough to prove the conversation is real (not echo).	power-user
`scenario`	No	Optional scenario nudge added to both personas' system prompts. Example: 'Topic: should the team switch from PostgreSQL to MongoDB? Have a real disagreement.' Leave empty to let the personas freestyle.
`smokeUrl`	No	URL of the smoke page on the LiveKit SFU droplet. Defaults to staging.	https://livekit-staging.comms.paradisemodern.com/smoke/
`paradiseBase`	No	Paradise Comms API base URL. Defaults to staging; pass production when ready.	https://comms.staging.paradisemodern.com
`paradiseToken`	Yes	Paradise Comms portfolio bearer token (e.g. paradise-staging_test_…). Get one by running scripts/seed-staging.ts in the paradisemodern repo, OR via POST /api/admin/comms/tokens as a super_admin.

Output Schema

TableJSON Schema

Name	Required	Description	Default
`result`	No	Tool result payload (JSON object)

TDQS

A4.3/5.0

Behavior5/5

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

Annotations already indicate non-read-only and non-idempotent behavior, but the description adds significant context: it's a real call, not simulation; each persona uses TTS and Whisper STT; the call goes through publisher/SFU/subscriber/recording/webhook; and it returns a transcript and recording hint. This fully explains the material side effects an agent should expect.

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?

The description is front-loaded with the core action and the crucial 'not simulation' caveat, and every sentence adds value. The phrase 'driven by PM Brain's realtime by default' is slightly vague for an agent without more context, but overall the prose is tight and well-ordered.

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?

Output schema exists, all parameters are documented, and annotations cover the safety profile. The description covers the end-to-end path, TTS/STT, recording, and demo purpose. Minor gaps: no estimated wall-clock duration, cost boundary, or explicit failure conditions, leaving a small amount to inference for a complex real-call tool.

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

Parameters3/5

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

Schema description coverage is 100%, and each parameter already has a detailed description including defaults, enums, and meaning. The description adds only light context around the personas, so the baseline of 3 applies — the schema bears the semantic load.

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?

The description is explicit and specific: it spawns two AI personas on a real LiveKit voice call and runs a structured conversation. It clearly distincts this from simulation ('this isn't simulation, it's a real WebRTC call with real audio') and from the sibling test-submission tools (submit_test, submit_test_batch) by emphasizing the live two-agent voice conversation.

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?

The description states a clear purpose: 'Used to verify Paradise Comms end-to-end (publisher → SFU → subscriber → recording → outbound webhook) and to demo agent-to-agent voice.' This tells the agent when to choose this tool, but it does not name alternatives or provide explicit when-not-to-use conditions.

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

All TestMyVibes tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A3.9/5.0

Disambiguation4/5

The tools cover a wide range of functionalities, but each has a clearly distinct purpose. For example, submit_test, submit_test_batch, submit_combo, and submit_interaction_scene are all different types of submissions with unique parameters. However, the sheer number of tools (43) might cause some initial confusion, but descriptors resolve ambiguity.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern in snake_case (e.g., list_projects, create_project, get_test_results). The only exception is 'whoami', which is a common idiom and does not break the pattern. Overall, naming is highly predictable.

Tool Count3/5

43 tools is on the high side for a single server. The domain is broad (testing, worker marketplace, credits, cards, feedback, video), so the count is justifiable. However, it borders on being overwhelming, and some tools could be consolidated (e.g., multiple submit_* variants).

Completeness3/5

The tool surface covers core workflows like project creation, test submission, result retrieval, worker management, and credit operations. However, there are gaps: no update or delete for projects, no delete for worker offerings, and no user-facing combo editing (though combos are predefined). These are minor but noticeable.

Resources

Need Help?