MCP Fish Server

fish_create_model

Destructive

Save a cloned voice as a reusable public model from authorized reference audio; use the returned ID as reference_id for audio generation. Others can discover and reuse it, with no private option.

Instructions

Save a cloned voice long-term and publicly; other users may discover and use it.

Use only reference audio authorized for processing and publication as a
reusable voice. There is no private/unlisted option. For one-shot reference
conditioning without saving a reusable model, use fish_generate_audio.
A successful create returns HTTP 201 and a voice _id; pass that _id as
reference_id to fish_generate_audio. Do not retry an uncertain create blindly.

Input Schema

Name	Required	Description
`tags`	No
`texts`	No
`title`	Yes	Name of the reusable public voice.
`voices`	Yes	One HTTP(S) URL to authorized reference audio; a string, not an array or upload.
`cover_image`	No
`description`	No
`generate_sample`	No
`enhance_audio_quality`	No

Output Schema

Name	Required	Description	Default
`result`	Yes

Schema Changelog

Changes observed during successful MCP inspections.

Addedv0.1.9Nov 2, 2026

TDQS

A4.4/5.0

Behavior5/5

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

Annotations already declare destructive=true and idempotent=false, but the description adds crucial non-annotation context: public discoverability by other users, the authorization requirement on reference audio, the HTTP 201 plus voice _id outcome, and the explicit warning not to blindly retry an uncertain create. This is exactly the additive behavioral detail the annotations cannot convey.

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

Conciseness5/5

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

The public-permanence constraint is front-loaded before alternatives and return details, and every sentence carries operational weight: authorization, no-private-option, sibling routing, return value, retry caution. No filler or restatement of the name.

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

Completeness3/5

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

Return semantics are partly covered and an output schema exists, so the 201/_id mention is sufficient there. However, for an 8-parameter destructive mutation with only 25% schema coverage, the description leaves the optional parameters entirely opaque, which is a meaningful completeness gap.

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?

Schema description coverage is only 25% (title and voices are the only documented properties), so the description must compensate for six undocumented parameters: tags, texts, cover_image, description, generate_sample, enhance_audio_quality. The description adds a semantic constraint on the 'voices' input (authorized reference audio) but says nothing about any of the remaining parameters, leaving the majority of the 8-param surface unexplained in both places.

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 opening line states a specific verb and resource ('Save a cloned voice') plus the defining scope ('long-term and publicly'), which immediately separates it from the transient generation sibling. An agent can classify the tool 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 Guidelines5/5

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

Explicitly names the alternative ('For one-shot reference conditioning without saving a reusable model, use fish_generate_audio') and the selection condition. It also states the hard constraint that there is no private/unlisted option, so an agent cannot mis-assume a hidden visibility setting.

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

Deploy Server