# TikiAPI (remote · mcp.tikiapi.com)

Images, video, voice and music from your AI assistant. Pay per result, no subscription.

- Trust score: 80/100 (high trust)
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-22

## Components

- remote · `mcp.tikiapi.com`: 80/100 (this document), [markdown](https://verifymcp.io/servers/com-tikiapi-mcp/mcp.md), [page](https://verifymcp.io/servers/com-tikiapi-mcp/mcp)

## Channel facts

- Endpoint: `https://mcp.tikiapi.com/mcp`
- Transports: `streamable-http`
- Auth: `required`
- Version: `1.0.0`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-11-22.

- **Endpoint Security**: 92/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token.
  - HTTPS is enforced; there's no plaintext access path.
  - HSTS check failed: the Strict-Transport-Security header is absent.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
  - The authorisation server supports Client ID Metadata Documents, the current MCP client-registration mechanism.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 72/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (good).
  - Context-footprint check failed: tool/resource definitions use about 4373 tokens (~242/item across 18 items; 18 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 17/100
  - Stability observed for 5 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 94/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 78% of tool parameters carry a description.
  - Structured output schemas are declared (100% of tools); any adoption earns full credit.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - All 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.
  - An AI judge read all 19 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a current MCP spec version (2026-07-28).

## Install

### How do I install the TikiAPI MCP server?

TikiAPI is a hosted endpoint at https://mcp.tikiapi.com/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add --transport http com-tikiapi-mcp 'https://mcp.tikiapi.com/mcp'
```

### Cursor

```json
{
  "mcpServers": {
    "com-tikiapi-mcp": {
      "url": "https://mcp.tikiapi.com/mcp"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "com-tikiapi-mcp": {
      "type": "http",
      "url": "https://mcp.tikiapi.com/mcp"
    }
  }
}
```

### Codex

```toml
[mcp_servers.com-tikiapi-mcp]
url = "https://mcp.tikiapi.com/mcp"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "com-tikiapi-mcp": {
      "type": "remote",
      "url": "https://mcp.tikiapi.com/mcp",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add com-tikiapi-mcp --url 'https://mcp.tikiapi.com/mcp' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  com-tikiapi-mcp:
    url: "https://mcp.tikiapi.com/mcp"
```

### Netclaw

```json
{
  "McpServers": {
    "com-tikiapi-mcp": {
      "Transport": "http",
      "Url": "https://mcp.tikiapi.com/mcp"
    }
  }
}
```

### Vellum

```bash
assistant mcp add com-tikiapi-mcp -t streamable-http -u 'https://mcp.tikiapi.com/mcp'
```

### Other

```json
{
  "mcpServers": {
    "com-tikiapi-mcp": {
      "type": "http",
      "url": "https://mcp.tikiapi.com/mcp"
    }
  }
}
```

The mcpServers block is a cross-client convention. Remote transports vary, so check your client's docs.

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-11-22 (score 80, +1)

No change was recorded against any check on this day. Stability & Change Management went from 13 to 17. That category is still filling its 30-day observation window: 4 days of observed history at the previous scan, 5 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-20 (score 79, +1)

- [functional] Server version: 6249797619fbe8931198ad527be6d5061a277d73 → 917b4243b10a8943693cf36146e9bc7f337e71e8

### 2026-11-19 (score 78, 0)

- [security] The server rewrote its instructions, which are the text every model session reads
- [security] Tool “edit_video” rewrote its description, which is the text the model reads
- [security] Tool “estimate_cost” rewrote its description, which is the text the model reads
- [security] Tool “generate” rewrote its description, which is the text the model reads
- [security] Tool “get_edit” rewrote its description, which is the text the model reads
- [security] Tool “get_scenario” rewrote its description, which is the text the model reads
- [security] Tool “list_models” rewrote its description, which is the text the model reads
- [security] Tool “speak” rewrote its description, which is the text the model reads
- [security] Tool “transcribe” rewrote its description, which is the text the model reads
- [functional regression] Tool coverage: 97% → 78%
- [functional] Server version: 5a851a05f40d370dead0c29386ce605c69907cf2 → 6249797619fbe8931198ad527be6d5061a277d73
- [cosmetic] “get_edit” added an optional parameter “wait_seconds”
- [cosmetic] “speak” added an optional parameter “run_again”
- [cosmetic] “transcribe” added an optional parameter “run_again”
- [cosmetic] “transcribe” added an optional parameter “words”
- [cosmetic] “edit_video” reworded the description of “captions”
- [cosmetic] “estimate_cost” reworded the description of “aspect_ratio”
- [cosmetic] “estimate_cost” reworded the description of “audio”
- [cosmetic] “estimate_cost” reworded the description of “audio_mode”
- [cosmetic] “estimate_cost” reworded the description of “audio_urls”
- [cosmetic] “estimate_cost” reworded the description of “character_id”
- [cosmetic] “estimate_cost” reworded the description of “character_ids”
- [cosmetic] “estimate_cost” reworded the description of “duration”
- [cosmetic] “estimate_cost” reworded the description of “fast”
- [cosmetic] “estimate_cost” reworded the description of “image_mode”
- [cosmetic] “estimate_cost” reworded the description of “image_urls”
- [cosmetic] “estimate_cost” reworded the description of “instrumental”
- [cosmetic] “estimate_cost” reworded the description of “lyrics”
- [cosmetic] “estimate_cost” reworded the description of “model”
- [cosmetic] “estimate_cost” reworded the description of “prompt”
- [cosmetic] “estimate_cost” reworded the description of “resolution”
- [cosmetic] “estimate_cost” reworded the description of “speaker_boxes”
- [cosmetic] “estimate_cost” reworded the description of “style”
- [cosmetic] “estimate_cost” reworded the description of “video_urls”

### 2026-11-18 (score 78, 0)

- [functional regression] Schema quality: 226 → 252
- [functional improvement] Stability: unverified → 0.03
- [functional] Server version: e24a8421a0341e7e7ed077859493338511b497b6 → 5a851a05f40d370dead0c29386ce605c69907cf2
- [cosmetic] “estimate_cost” added an optional parameter “audio_mode”
- [cosmetic] “estimate_cost” added an optional parameter “fast”
- [cosmetic] “estimate_cost” added an optional parameter “speaker_boxes”
- [cosmetic] “generate” added an optional parameter “audio_mode”
- [cosmetic] “generate” added an optional parameter “fast”
- [cosmetic] “generate” added an optional parameter “speaker_boxes”
- [cosmetic] “estimate_cost” reworded the description of “audio_urls”
- [cosmetic] “generate” reworded the description of “audio_urls”

### 2026-11-17 (score 78, +41)

- [security improvement] Transport: unverified → pass
- [security improvement] Injection markers: unverified → pass
- [security] First check of Judged manipulation: pass
- [security] Authorization: Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token.
- [functional improvement] MCP protocol: unverified → pass
- [functional improvement] Schema quality: unverified → 100
- [functional improvement] Tool coverage: unverified → 100
- [functional] First check of Schema quality: fail
- [functional] First check of Tool coverage: 96
- [functional] First check of Schema quality: fail
- [functional] First check of Destructive annotations: 100
- [functional] First check of Schema quality: good
- [functional] First check of Tool coverage: 100

### 2026-11-16 (score 37)

First indexed and scored.

## MCP tools (18)

### `list_models` (~110 tokens)

List models

Models with their limits and prices in USD: the public catalog before sign-in, this account's own after. Free. Empty fields are left out: no max_reference_images means the model takes no reference images. Text (chat) models are left out unless kind is text: they run over REST (POST /v1/chat/completions), not through these tools.

Input parameters:

- `kind` (string): Only models of this kind: video, image, audio (music, sound effects, speech) or text.

Output parameters:

- `data` (array)
- `object` (string)

### `estimate_cost` (~212 tokens)

Estimate cost

The price of an order before placing it: the same fields, the same checks and the same price as generate (see its fields). Nothing is made and nothing is charged. Before sign-in: the public price (reference videos, audio and characters need sign-in, since their seconds and photos count); after: this account's price and balance.

Input parameters:

- `aspect_ratio` (string)
- `audio` (boolean)
- `audio_mode` (string)
- `audio_urls` (array)
- `character_id` (string)
- `character_ids` (array)
- `duration` (integer)
- `fast` (boolean)
- `image_mode` (string)
- `image_urls` (array)
- `instrumental` (boolean)
- `lyrics` (string)
- `model` (string, required)
- `prompt` (string)
- `resolution` (string)
- `speaker_boxes` (array)
- `style` (string)
- `video_urls` (array)

Output parameters:

- `available_balance` (string): US dollars, for example 0.84
- `can_order_now` (boolean)
- `charge` (string): With a key or sign-in: what the order would take from the balance (0 for a test key).
- `enough_balance` (boolean)
- `kind` (string)
- `list_price` (string): A promo price: the price after promo_until.
- `model` (string)
- `object` (string)
- `price` (string): What this order costs
- `promo_until` (string): A promo price: when it ends (ISO, UTC). An order placed before then keeps this price.

### `upload_media` (~360 tokens)

Upload media

Stores a file of your own (PNG, JPEG or WebP image, MP4 or MOV video, MP3 or WAV audio) and returns a link and an id, either one for image_urls, video_urls or audio_urls (and create_character, edit_video). Three ways in: data — the file as base64 here (up to 20 MB); prepare — a signed link to PUT the file to yourself, with a ready curl command (videos up to 50 MB; best for big files, no base64 in the chat); url — we fetch a public link. Kept 48 hours, then deleted.

Input parameters:

- `bytes` (integer): mode prepare: the file's exact size in bytes (for example: wc -c < file). The link takes exactly this many.
- `content_type` (string): mode prepare: the file's type, for example image/png, video/mp4, audio/mpeg. Other modes read the type from the file itself.
- `data` (string): mode data: the file, base64-encoded (a data: URL works too).
- `filename` (string): The file's name; in mode prepare its extension gives the type when content_type is left out.
- `mode` (string): How the file comes. data (default): base64 here, up to 20 MB. prepare: a signed link you PUT the file to yourself (a ready curl command comes back): videos up to 50 MB, images 20 MB, audio 15 MB — be…
- `url` (string): mode url: a public https link to the file itself.

Output parameters:

- `bytes` (number)
- `content_type` (string)
- `expires_at` (string)
- `id` (string)
- `kind` (string)
- `object` (string)
- `url` (string)

### `list_generations` (~76 tokens)

List generations

This account's orders, newest first, with their status, price and result links. Pass next as before for older ones.

Input parameters:

- `before` (string): An order id: list the ones before it (the next value of the previous page).
- `limit` (integer): How many, 1–100 (default 20).

Output parameters:

- `data` (array)
- `next` (string|null)
- `object` (string)

### `generate` (~961 tokens)

Generate

Makes an image, a video, a song or a sound effect and waits for it (up to about a minute; a video or a song usually takes longer). A music order gives two takes of the song. Spends the account balance; a failed order costs nothing. The answer links each original and shows a small preview (a picture at 1024 px, a clip as 4 frames: look for a white flash or grey blacks at the start). If it is still running, the answer has eta_seconds and poll_after_seconds: check with get_generation then — don't start it again. wait false places it and answers at once with its id: for several pictures at once, collected with one wait_for_generations. Videos: one at a time, each looked at before the next. The same request again within 10 minutes returns the same order (repeated: true), never a second charge: for another variant pass run_again true.

Input parameters:

- `aspect_ratio` (string): For example 1:1, 16:9, 9:16.
- `audio` (boolean): Video: false for a silent clip, where the model allows it (audio_optional in list_models).
- `audio_mode` (string): Lip sync (lip_sync in list_models) with two people in one photo and two audio files, one per person (the first is the person on the left). parallel: both tracks share one timeline and are the same le…
- `audio_urls` (array): Reference audio (MP3/WAV): public https links, or upload_media links or ids. Video models only, free; most need an image or a video next to it. Lip-sync models (lip_sync in list_models): the speech t…
- `character_id` (string): A saved person (create_character): their photos go first among the references and the prompt opens with an identity block. Don't describe their face or hair in the prompt; put the framing first (clos…
- `character_ids` (array): Two saved people in one order, instead of character_id.
- `duration` (integer): Seconds: a video's length; a track's (music, optional: the model picks otherwise); a sound effect's (default 5, priced by the second).
- `fast` (boolean): The model's faster mode at the same price, where list_models shows fast_mode (LongCat: about 10% quicker, the picture may change a little).
- `image_mode` (string): Video: how image_urls are used. reference (default) — things to bring in; first_frame — the clip opens on the one image; first_last_frame — opens on the first, ends on the second.
- `image_urls` (array): Reference images: public https links, or upload_media links or ids.
- `instrumental` (boolean): Music: true for no vocals.
- `lyrics` (string): Music: the words to sing, as written (sections like [Verse], [Chorus] help). Leave out for the model to write them from the prompt.
- `model` (string, required): Model id from list_models.
- `prompt` (string, required): What to make.
- `resolution` (string): For example 1K, 2K, 4K for images; 720p for video.
- `run_again` (boolean): Make another one even if the same request was just made.
- `speaker_boxes` (array): Lip sync with two speakers, optional: where each person is, [x_min, y_min, x_max, y_max] in the photo's pixels, in the order of audio_urls. Left out: the first track is the person on the left, the se…
- `style` (string): Music: genre, mood, instruments, voice, for example lo-fi hip hop, warm Rhodes, female vocal.
- `video_urls` (array): Reference videos (MP4/MOV): public https links, or upload_media links or ids. Video models only; their seconds are billed too.
- `wait` (boolean): true (default): wait for the result, up to about a minute. false: answer at once with the order's id; collect it with wait_for_generations or get_generation.

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `error`: Why it failed, or null.
- `eta_seconds` (number|null): While it runs: about how many seconds are left.
- `id` (string)
- `message` (string)
- `model` (string)
- `object` (string)
- `outputs` (array): The results: links on our CDN.
- `poll_after_seconds` (number): While it runs: when to check again with get_generation.
- `price` (string): US dollars, for example 0.84
- `repeat_note` (string)
- `repeated` (boolean): true: this is an order placed a few minutes ago with the same request, not a new one (see repeat_note).
- `status` (string): queued, running, succeeded, failed or canceled

### `get_generation` (~55 tokens)

Get generation

An order's status and result: links to the originals and a small preview to look at. While it runs: eta_seconds (likely time left) and poll_after_seconds (when to check again).

Input parameters:

- `id` (string, required)

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `error`: Why it failed, or null.
- `eta_seconds` (number|null): While it runs: about how many seconds are left.
- `id` (string)
- `message` (string)
- `model` (string)
- `object` (string)
- `outputs` (array): The results: links on our CDN.
- `poll_after_seconds` (number): While it runs: when to check again with get_generation.
- `price` (string): US dollars, for example 0.84
- `repeat_note` (string)
- `repeated` (boolean): true: this is an order placed a few minutes ago with the same request, not a new one (see repeat_note).
- `status` (string): queued, running, succeeded, failed or canceled

### `get_account` (~18 tokens)

Get account

The balance in USD and where to top up.

Output parameters:

- `balance` (object)
- `id` (string)
- `name` (string)
- `object` (string)
- `top_up_url` (string)

### `speak` (~248 tokens)

Speak

Reads text aloud and returns a link to the MP3 (kept 30 days). voice is a free-text direction, e.g. "a warm, amused British narrator"; a [bracketed] cue before a sentence sets its emotion, e.g. "[whispering] It's here.". To clone a voice, give reference_audio_url (MP3 or WAV, 10–30 s of clean speech) and, ideally, reference_text. Billed per character read (the direction too); see list_models. The same request again within 10 minutes returns the same recording, not a second charge.

Input parameters:

- `model` (string): A speech model from list_models (kind audio); the first one when left out.
- `reference_audio_url` (string): A recording to clone the voice from: public https link or an upload_media link.
- `reference_text` (string): What is said in that recording.
- `run_again` (boolean): Make another take even if the same request was just made.
- `text` (string, required): What to say. May contain [bracketed] emotion cues.
- `voice` (string): How it sounds: a free-text voice direction.

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `id` (string)
- `object` (string)
- `price` (string): US dollars, for example 0.84
- `repeated` (boolean)
- `status` (string): succeeded, or running: then call speak again with the same arguments (the same call, no new charge).
- `url` (string|null)

### `transcribe` (~302 tokens)

Transcribe

Writes down what is said in a recording, with each segment's start and end in seconds. Give url (a public https link or an upload_media link) or data (the file as base64, up to 20 MB). MP3, WAV, FLAC, OGG, M4A, MP4 or MOV. For captions, pass the answer's id to edit_video as captions { transcription_id } (kept 48 hours); words true also returns every word's timing here. Billed per minute of the recording as we measure it; see list_models. The same request again within 10 minutes is the same call, not a second charge.

Input parameters:

- `data` (string): Or the file itself, base64-encoded (a data: URL works too).
- `language` (string): ISO-639-1 code, e.g. en; detected when left out.
- `model` (string): A transcription model from list_models (kind audio); the first one when left out.
- `prompt` (string): Spelling hints: names and terms that occur in the recording.
- `run_again` (boolean): Transcribe it anew even if the same request was just made.
- `url` (string): The recording: a public https link or an upload_media link.
- `words` (boolean): true: also return every word's start and end (long for a long recording). Default false: captions take the id instead.

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `id` (string): Give it to edit_video as captions { transcription_id }.
- `price` (string): US dollars, for example 0.84
- `repeated` (boolean)
- `status` (string): succeeded, or running: then call transcribe again with the same arguments (the same call, no new charge).
- `text` (string)

### `edit_video` (~720 tokens)

Edit video

Edits videos on our servers. reframe: one video to 9:16, 16:9 or 1:1, cropped around a focus point or padded with a blurred copy of itself. assemble: clips joined in order (each optionally trimmed), with an optional audio track over them (music or a voice-over, with its volume and the clips' own volume), optional burned-in captions from a transcribe call (by its id), and loudness brought to -14 LUFS; 1080x1920 or the first clip's size. Inputs: links to generation results (our CDN) or links from upload_media. $0.002 per second of output: the most it can cost is held, the real length is charged; a failed edit costs nothing. Limits: 20 clips of up to 120 s, 180 s of output. Waits about a minute; if still running, check with get_edit (wait_seconds holds until it is done) — don't start it again. The answer shows a sheet of the cut's frames to look at.

Input parameters:

- `aspect_ratio` (string): reframe: the new aspect ratio (required for reframe).
- `audio_start` (number): assemble: where in the track to start, seconds.
- `audio_url` (string): assemble: an audio track (MP3/WAV/M4A) laid over the clips, cut or padded to their length.
- `audio_volume` (number): assemble: the track's volume, 1 = as is (default).
- `captions` (object): assemble: burned-in captions. Best: { transcription_id: "<id from transcribe>" } (the voice track's transcript, kept 48 hours). Or the timed words themselves: { words: [{ text, start, end }] } or { s…
- `clip_volume` (number): assemble: the clips' own sound, 1 = as recorded (default), 0 = muted.
- `clips` (array): assemble: the clips in order; start and end (seconds) trim each one.
- `end` (number): reframe: end of the part to keep, seconds.
- `fit` (string): assemble: how clips of another shape fill the frame; blur (default) keeps them whole.
- `focus_x` (number): reframe, crop: what to keep in view, 0 (left) to 1 (right); default 0.5.
- `focus_y` (number): reframe, crop: 0 (top) to 1 (bottom); default 0.5.
- `mode` (string): reframe: crop (default) fills the frame; blur keeps the whole picture on a blurred background.
- `operation` (string, required)
- `resolution` (string): reframe: output size, default 1080p (1080x1920 for 9:16).
- `run_again` (boolean): Make another one even if the same request was just made.
- `size` (string): assemble: 1080x1920 (default) or the first clip's own size.
- `start` (number): reframe: start of the part to keep, seconds.
- `video_url` (string): reframe: the video to reframe.

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `id` (string)
- `object` (string)
- `operation` (string)
- `price`
- `repeated` (boolean): true: this is an edit placed a few minutes ago with the same request, not a new one.
- `status` (string)

### `get_edit` (~81 tokens)

Get edit

An edit's status and result (from edit_video): a link to the video and a sheet of its frames to look at. wait_seconds holds the call until it is done.

Input parameters:

- `id` (string, required)
- `wait_seconds` (integer): Wait up to this long for a running edit, 0–50 s (default 0: just look).

Output parameters:

- `charged` (string): US dollars, for example 0.84
- `id` (string)
- `object` (string)
- `operation` (string)
- `price`
- `repeated` (boolean): true: this is an edit placed a few minutes ago with the same request, not a new one.
- `status` (string)

### `wait_for_generations` (~136 tokens)

Wait for generations

Waits for several orders at once (up to 12 ids, up to 50 s) and returns the finished ones with their results and the ones still running. Use it after generate with wait false, e.g. four pictures placed one after another, then one wait for all of them. If some are still running, call it again with their ids after poll_after_seconds; never order them again.

Input parameters:

- `ids` (array, required): Order ids from generate or list_generations.
- `timeout_seconds` (integer): How long to wait at most, 0–50 s (default 50); 0 just looks.

### `list_scenarios` (~81 tokens)

List scenarios

Ready-made plans (an ad video end to end, an ad script, a consistent character, a voice-over, a product shoot, a thumbnail and more) with the fields each one asks for. After sign-in, also the skills the person imported into their TikiAPI account: named my-…, marked yours. Free, no sign-in needed.

Output parameters:

- `data` (array)
- `object` (string)

### `get_scenario` (~165 tokens)

Get scenario

A scenario's worked plan with your fields filled in, for example name "ad-video" with fields { "product": "a reusable coffee cup" }. Follow it step by step: it says where to stop for the person's yes before anything is paid. The models it names are the ones on sale now. The person's own imported skills (my-…, after sign-in) come back as their own instructions; a file that comes with one is name "my-<skill>/<path>". Free, no sign-in needed.

Input parameters:

- `fields` (object): Its fields as text, by name (list_scenarios shows them).
- `name` (string, required): The scenario, from list_scenarios; for a file of an imported skill, my-<skill>/<path>.

Output parameters:

- `name` (string)
- `object` (string)
- `text` (string): The worked plan to follow.
- `title` (string)

### `create_character` (~199 tokens)

Create character

Saves a person to put into generate by character_id, so their face holds from picture to picture and in video. Give a clean close-up of the face and, separately, a full-length photo — never a front/side/back sheet, which reads as several people. notes (outfit and details) open every prompt with this person. Kept until deleted.

Input parameters:

- `body_image` (string): A separate full-length photo: a public https link, an upload_media link, or an upload id.
- `extra_images` (array): Up to 3 more photos of the same person.
- `face_image` (string, required): A clean close-up of the face: a public https link, an upload_media link, or an upload id.
- `name` (string, required): What to call them; the prompt names them by it.
- `notes` (string): Outfit and details, for example: black leather jacket, silver ring on the left hand.

Output parameters:

- `id` (string)
- `name` (string)
- `object` (string)

### `list_characters` (~31 tokens)

List characters

This account's saved people, newest first, with their photos (links work for an hour) and notes.

Output parameters:

- `data` (array)
- `object` (string)

### `delete_character` (~34 tokens)

Delete character

Deletes a saved person and their photos. Orders already made keep their results.

Input parameters:

- `id` (string, required): The character's id.

Output parameters:

- `deleted` (boolean)
- `id` (string)
- `object` (string)

### `report_problem` (~137 tokens)

Report a problem

Tells the TikiAPI team about a problem: an order stuck for much longer than its eta_seconds, a charge that looks wrong, a broken result. Give the ids it concerns (orders, edits, speech or transcription calls) and a short description. Use it once per problem, after checking get_generation: a failed order is never charged.

Input parameters:

- `ids` (array): The ids it concerns: orders, edits, speech or transcription calls.
- `problem` (string, required): stuck_order, odd_charge, bad_result or other.
- `text` (string, required): What happened, in a few sentences: what was expected and what came back.

Output parameters:

- `id` (string)
- `message` (string)
- `object` (string)
- `received` (boolean)

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/com-tikiapi-mcp/mcp#diagnostics

## Score history

- 2026-11-22: 80
- 2026-11-21: 79
- 2026-11-20: 79
- 2026-11-19: 78
- 2026-11-18: 78
- 2026-11-17: 78
- 2026-11-16: 37

## Common questions

### What is the TikiAPI MCP server?

TikiAPI is an MCP server listed in the public MCP registry as com.tikiapi/mcp. Images, video, voice and music from your AI assistant. Pay per result, no subscription. This page covers its hosted endpoint (https://mcp.tikiapi.com/mcp).

### Is the TikiAPI MCP server safe to use?

TikiAPI scores 80 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the TikiAPI MCP server expose?

TikiAPI exposes 18 tools: list_models, estimate_cost, upload_media, list_generations, generate, and 13 more. Their descriptions and schemas cost roughly 3,926 tokens of context every time the server is loaded.

### Does the TikiAPI MCP server require authentication?

Yes. TikiAPI asked us for credentials when we connected, so you will need to authorise it in your MCP client before it can do anything.

### Is the TikiAPI MCP server still maintained?

TikiAPI is still listed as active in the MCP registry. We last reached this channel on 22 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What did VerifyMCP test on the TikiAPI MCP server?

VerifyMCP ran 13 checks on TikiAPI, covering TLS certificate, Authorization, HTTPS, HSTS header and 9 more. Supply Chain Security and Provenance & Transparency do not apply to a hosted endpoint, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.

## Links

- Remote endpoint: https://mcp.tikiapi.com/mcp
- Website: https://tikiapi.com/mcp
- Changelog RSS feed: https://verifymcp.io/servers/com-tikiapi-mcp/mcp.xml
- Changelog JSON feed: https://verifymcp.io/servers/com-tikiapi-mcp/mcp/changelog.json
- Trust report (JSON): https://verifymcp.io/servers/com-tikiapi-mcp/mcp.json
- HTML version of this page: https://verifymcp.io/servers/com-tikiapi-mcp/mcp
