Agent teams, swarms, memory and hooks for Claude Code and Codex, served over MCP.
Bedrock Multi-Model is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
What it does
Ask, compare and judge 35 Bedrock text models, generate and edit images, plus video and embeddings.
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user bedrock-multi-model-mcp '--' npx -y bedrock-multi-model-mcp
# 2 — ask your agent for something these tools do
# bedrock_ask, bedrock_compare, bedrock_judge and 6 more
Bedrock Multi-Model costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Bedrock Multi-Model as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
This is a public server: you run it yourself and this marketplace is not in the path. You need Node.js — the current LTS (24) is the safe choice; npx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user bedrock-multi-model-mcp '--' npx -y bedrock-multi-model-mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
Nothing is required, but it can take AWS_BEARER_TOKEN_BEDROCK (Bedrock API key. Leave unset to use the standard AWS credential chain.), AWS_REGION (Region for text and embedding calls (default us-east-1).), AWS_PROFILE (Named AWS profile, when using IAM credentials instead of a bearer token.), BEDROCK_MCP_OUTPUT_DIR (Directory for generated images (default: cwd, or ~/bedrock-images).) if you want to set them.
One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.
9 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Read from its source code. A server that builds its tool list when it starts can list more tools than this.
Send a prompt to any AWS Bedrock model via the Converse API.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| model | string | yes | Model alias (e.g. 'claude-opus', 'kimi', 'mistral-large', 'nova-pro', 'deepseek') or full Bedrock model ID |
| prompt | string | yes | The user prompt to send to the model |
| system | string | no | Optional system prompt |
| temperature | number | no | Sampling temperature (0-1, default 0.7). Ignored by models that don't support it. |
| max_tokens | integer | no | Maximum tokens to generate |
| effort | string | no | Thinking depth and token spend for models that take it (Claude except Haiku 4.5, GPT-6.1, Grok 4.7, Kimi K3, Gemma 4 up to xhigh): low, medium, high, xhigh, max. Claude Opus 5.5 and Haiku 5.5 default |
| schema | object | no | JSON schema the reply must match (structured output). Object schemas need additionalProperties: false. Supported by Claude, GPT-6.1, GPT-OSS, Grok 4.7, Kimi K3, Gemma 4, Mistral Large 3, Ministral 3, |
Send the same prompt to multiple Bedrock models and compare responses side by side.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| models | array | yes | List of 2-5 model aliases or IDs to compare (e.g. ['kimi', 'nova-pro', 'mistral-large']) |
| prompt | string | yes | The prompt to send to all models |
| system | string | no | Optional system prompt sent to all models |
| temperature | number | no | Sampling temperature for all models |
| max_tokens | integer | no | Maximum tokens per model response |
| effort | string | no | Thinking depth and token spend for models that take it (Claude except Haiku 4.5, GPT-6.1, Grok 4.7, Kimi K3, Gemma 4 up to xhigh): low, medium, high, xhigh, max. Claude Opus 5.5 and Haiku 5.5 default |
| schema | object | no | JSON schema the reply must match (structured output). Object schemas need additionalProperties: false. Supported by Claude, GPT-6.1, GPT-OSS, Grok 4.7, Kimi K3, Gemma 4, Mistral Large 3, Ministral 3, |
Send the same prompt to 2-5 Bedrock models, then have a judge model (default Claude Opus)
| Parameter | Type | Required | Notes |
|---|---|---|---|
| models | array | yes | List of 2-5 candidate model aliases or IDs (e.g. ['kimi', 'nova-pro', 'deepseek']) |
| prompt | string | yes | The prompt every candidate answers |
| judge | string | no | Judge model alias or ID (default 'claude-opus') |
| criteria | string | no | What the judge scores on (default 'correctness, completeness, clarity and concision') |
| synthesize | boolean | no | Also return one merged final answer (default false) |
| system | string | no | Optional system prompt sent to the candidates |
| temperature | number | no | Sampling temperature for the candidates |
| max_tokens | integer | no | Maximum tokens per candidate response |
| effort | string | no | Thinking depth and token spend for models that take it (Claude except Haiku 4.5, GPT-6.1, Grok 4.7, Kimi K3, Gemma 4 up to xhigh): low, medium, high, xhigh, max. Claude Opus 5.5 and Haiku 5.5 default |
List available Bedrock text models with their aliases, providers, use cases, and capabilities.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| provider | string | no | Filter by provider (e.g. 'Anthropic', 'OpenAI', 'xAI', 'Mistral', 'Amazon', 'Qwen', 'DeepSeek') |
Generate an image from a text prompt using AWS Bedrock image models
| Parameter | Type | Required | Notes |
|---|---|---|---|
| model | string | no | Image model alias: 'stable-ultra' (default, best quality), 'sd3.5', 'stable-core' (fastest/cheapest), or a full model ID |
| prompt | string | yes | Text description of the image to generate |
| negative_prompt | string | no | What to exclude from the image (e.g. 'blurry, low quality') |
| width | number | no | Desired width (default 1024). Stability models take an aspect ratio, not pixel dimensions: width and height only pick the closest supported ratio (1:1, 16:9, 9:16, 3:2, 2:3, 4:5, 5:4, 21:9, 9:21) and |
| height | number | no | Desired height (default 1024). See width - only the resulting aspect ratio is sent. |
| seed | integer | no | Random seed for reproducible results |
| output_dir | string | no | Directory to save the image. Absolute or relative to cwd. Auto-created if missing. Overrides BEDROCK_MCP_OUTPUT_DIR and default resolution. |
Edit or transform a local image with Stability AI Image Services on AWS Bedrock: fill or erase a masked
| Parameter | Type | Required | Notes |
|---|---|---|---|
| operation | string | yes | What to do: … |
| image_path | string | yes | Input image (PNG, JPEG or WebP), absolute or relative to cwd. For style-transfer, the image to restyle. |
| prompt | string | no | What the output should show |
| negative_prompt | string | no | What to keep out of the output |
| mask_path | string | no | inpaint, erase: black-and-white mask image, white where the image should change. Without it, the input image's alpha channel is used as the mask. |
| style_image_path | string | no | style-transfer: the image whose style is applied |
| search_prompt | string | no | search-replace: the object to find and replace |
| select_prompt | string | no | search-recolor: the object to find and recolor |
| left | integer | no | outpaint: pixels to add on the left |
| right | integer | no | outpaint: pixels to add on the right |
| up | integer | no | outpaint: pixels to add on top |
| down | integer | no | outpaint: pixels to add at the bottom |
| creativity | number | no | upscale-conservative and upscale-creative (0.1-0.5), outpaint (0.1-1): how much new detail to invent |
| control_strength | number | no | control-sketch, control-structure: how closely the output follows the input (default 0.7) |
| grow_mask | integer | no | inpaint, erase, search-replace, search-recolor: pixels to grow and soften the mask edge (default 5) |
| fidelity | number | no | style-guide: how closely the output follows the input image's style (default 0.5) |
| aspect_ratio | string | no | style-guide: output aspect ratio (default 1:1) |
| composition_fidelity | number | no | style-transfer: how closely the output keeps the input's composition (default 0.9) |
| style_strength | number | no | style-transfer: how strongly the style image applies (default 1) |
| change_strength | number | no | style-transfer: how much the input may change (default 0.9) |
Start a video generation job using Luma Ray 2 on AWS Bedrock.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| model | string | no | Video model alias: 'luma-ray' / 'ray2' (default), or a full model ID |
| prompt | string | yes | Text description of the video to generate |
| s3_uri | string | yes | S3 URI for output (e.g. 's3://my-bucket/videos/'). Bucket must be in the model's region. |
| duration_seconds | no | Video duration in seconds: 5 (default) or 9. Luma Ray 2 supports no other value. |
Check the status of a Bedrock video generation job (Luma Ray 2).
| Parameter | Type | Required | Notes |
|---|---|---|---|
| invocation_arn | string | yes | The job ARN returned by bedrock_generate_video |
Embed 2 or more texts with an AWS Bedrock embedding model (Amazon Titan, Cohere) and
| Parameter | Type | Required | Notes |
|---|---|---|---|
| texts | array | yes | List of 2-50 texts to embed and compare pairwise |
| model | string | no | Embedding model alias: 'titan-v2' (default), 'titan-multimodal', 'cohere' (Embed v4), or a full model ID |
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under MIT
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install Bedrock Multi-Model or to call it, at any volume.
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | moderate | capability blast radius (moderate) — client exposure if the model is manipulated | −3 |
| verification-discount | source | publisher verification (provenance) — cryptographic build provenance ties the artifact to its source | −0 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
Tools were statically extracted from the published source (9 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. moderate is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| bedrock_ask | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_compare | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_judge | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_list_models | no tags | |
| bedrock_generate_image | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_edit_image | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_generate_video | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| bedrock_video_status | no tags | |
| bedrock_embed_similarity | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.
Supply chain and provenance
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
Always latest — the install command names no version, so it installs the newest release when it runs; the current release on file is 0.8.0. Pin a release below and the command asks for exactly that one.
No release note was published with this version.
- The version this listing was on when mcprush began following its releases. The registry was not asked when it shipped, so this row carries no date.
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Any signed-in account that does not publish Bedrock Multi-Model can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.