Rawa

Remove a background

remove_background

Cut the background out of a finished image, leaving a transparent PNG of the subject. The source is a creative of the brand (creativeId) or an image on the web (imageUrl: a direct, public URL to a JPEG, PNG or WebP file). An image the user pasted or attached in the chat has no such URL and you cannot read it yourself: call request_image_upload right away (do not ask them to re-attach it), and once the image has landed use its creativeId here. An image you hold as a file on your own filesystem (a coding agent in a project) goes through upload_image_file instead. A linked image is added to the brand's library first and its creativeId is returned as sourceCreativeId; an image the library already holds is not added again, and sourceCreativeId is that creative, with alreadyInLibrary: true. Reuse sourceCreativeId for any further work on the same image instead of passing imageUrl again. Cost: 30 credits. Call with dryRun: true first and tell the user the cost. Once they agree, pass the quoted credits as expectedCredits; a call that would cost a different amount is refused. Give each new generation a new requestId, and repeat it unchanged only on a retry: a retry within 2 minutes returns the first result and is not charged again. Without a requestId, the same arguments within 2 minutes count as a retry. Needs generation allowed over MCP (see get_workspace); a dry run still quotes when it is not, and says why the call would be refused in wouldRefuse. Returns a jobId, creditsCharged and creditsRemaining; ready in 1 to 5 minutes. If a job panel shows, it tracks the work: do not poll, end your turn with one line saying it is under way, and later use show_creatives with the creativeIds you are told. Otherwise call wait_for_jobs with the jobIds until allFinal is true.

Input Schema

Name	Required	Description
`dryRun`	No	true: return the exact cost (credits, breakdown, balanceAfter) without charging or starting anything, and wouldRefuse when the organization's spending controls would refuse the real call.
`brandId`	No	A brand id from get_workspace or list_brands, or the brand's name. Omit it when the organization has a single brand.
`imageUrl`	No	A direct, publicly reachable https URL of the image to work on, when it is not in the library yet.
`requestId`	No	Your own unique id for this request. Use a new one for every new request, even with the same arguments. Repeat it unchanged only to retry this exact call within 2 minutes; the retry returns the first result.
`creativeId`	No	A creative id from list_creatives, search_library, get_job or get_upload.
`expectedCredits`	No	The credits the dryRun quoted and the user agreed to. The call is refused if it would cost a different amount.

Output Schema

Name	Required	Description	Default
No arguments

Schema Changelog

Changes observed during successful MCP inspections.

Changed2 schema fields changedOct 29, 2026

changedInput schema / properties / brandId / description

Previous value: -"A brand id from get_workspace or list_brands. Omit it when the organization has a single brand."New value: +"A brand id from get_workspace or list_brands, or the brand's name. Omit it when the organization has a single brand."

changedOutput schema / anyOf

Previous value: -[
-  {
-    "additionalProperties": false,
-    "properties": {
-      "balanceAfter": {
-        "type": "number"
-      },
-      "breakdown": {
-        "type": "string"
-      },
-      "credits": {
-        "type": "number"
-      },
-      "deduplicated": {
-        "const": true,
-        "description": "The same call was made in the last 2 minutes: this is its result, not charged again.",
-        "type": "boolean"
-      },
-      "dryRun": {
-        "const": true,
-        "type": "boolean"
-      },
-      "wouldRefuse": {
-        "description": "Set when the organization's MCP spending controls would refuse the real call: the refusal it would get.",
-        "type": "string"
-      }
-    },
-    "required": [
-      "dryRun",
-      "credits",
-      "breakdown",
-      "balanceAfter"
-    ],
-    "type": "object"
-  },
-  {
-    "additionalProperties": false,
-    "properties": {
-      "creditsCharged": {
-        "type": "number"
-      },
-      "creditsRemaining": {
-        "type": "number"
-      },
-      "deduplicated": {
-        "const": true,
-        "description": "The same call was made in the last 2 minutes: this is its result, not charged again.",
-        "type": "boolean"
-      },
-      "jobId": {
-        "type": "string"
-      },
-      "sourceCreativeId": {
-        "type": "string"
-      },
-      "watchToken": {
-        "type": "string"
-      }
-    },
-    "required": [
-      "jobId",
-      "watchToken",
-      "creditsCharged",
-      "creditsRemaining",
-      "sourceCreativeId"
-    ],
-    "type": "object"
-  }
-]New value: +[
+  {
+    "additionalProperties": false,
+    "properties": {
+      "balanceAfter": {
+        "type": "number"
+      },
+      "breakdown": {
+        "type": "string"
+      },
+      "credits": {
+        "type": "number"
+      },
+      "deduplicated": {
+        "const": true,
+        "description": "The same call was made in the last 2 minutes: this is its result, not charged again.",
+        "type": "boolean"
+      },
+      "dryRun": {
+        "const": true,
+        "type": "boolean"
+      },
+      "wouldRefuse": {
+        "description": "Set when the organization's MCP spending controls would refuse the real call: the refusal it would get.",
+        "type": "string"
+      }
+    },
+    "required": [
+      "dryRun",
+      "credits",
+      "breakdown",
+      "balanceAfter"
+    ],
+    "type": "object"
+  },
+  {
+    "additionalProperties": false,
+    "properties": {
+      "alreadyInLibrary": {
+        "const": true,
+        "description": "The brand's library already held this image, so no new creative was added: the id names that one.",
+        "type": "boolean"
+      },
+      "creditsCharged": {
+        "type": "number"
+      },
+      "creditsRemaining": {
+        "type": "number"
+      },
+      "deduplicated": {
+        "const": true,
+        "description": "The same call was made in the last 2 minutes: this is its result, not charged again.",
+        "type": "boolean"
+      },
+      "jobId": {
+        "type": "string"
+      },
+      "sourceCreativeId": {
+        "type": "string"
+      },
+      "watchToken": {
+        "type": "string"
+      }
+    },
+    "required": [
+      "jobId",
+      "watchToken",
+      "creditsCharged",
+      "creditsRemaining",
+      "sourceCreativeId"
+    ],
+    "type": "object"
+  }
+]

First observedOct 23, 2026

TDQS

A4.8/5.0

Behavior5/5

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

Annotations already declare readOnlyHint=false, openWorldHint=true, idempotentHint=false, destructiveHint=false, but the description adds substantial operational context: cost of 30 credits, dryRun quoting, expectedCredits refusal guard, requestId retry window, generation-allowed prerequisite, returned jobId/credits, 1–5 minute readiness, and job-panel versus wait_for_jobs behavior. It does not contradict the annotations.

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 long but front-loaded and dense with operational rules rather than filler. Every major sentence carries a distinct instruction, though the volume is high for a single description and could be tightened slightly without losing meaning.

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

Completeness5/5

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

For a complex mutation with cost, credit guards, source-routing alternatives, and asynchronous job handling, the description is complete. Since an output schema exists, it appropriately does not spend space explaining every return field, but still mentions jobId, creditsCharged, creditsRemaining, and sourceCreativeId metadata needed to continue the workflow.

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 100%, so the schema already documents each parameter. The description still adds meaningful cross-parameter semantics beyond the schema: imageUrl must be a direct public JPEG/PNG/WebP URL, requestId must be new per generation and repeated only on retry, and expectedCredits must match the quoted dry-run amount to avoid refusal.

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+resource: cut the background out of a finished image, leaving a transparent PNG. It explicitly distinguishes among source types (brand creative, public image URL, chat attachment, local filesystem) and names the sibling tools to use for each, so an agent can route correctly without opening another 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?

Provides explicit when-to-use guidance and alternatives: use request_image_upload for pasted/attached chat images, upload_image_file for local filesystem images, and reuse sourceCreativeId instead of imageUrl for already-linked images. It also prescribes the dry-run-then-confirm workflow and the retry/idempotency behavior, leaving little to inference.

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

All Rawa tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?