Cloud Agent Settings | Creor

Configure how cloud agents run in your workspace. These settings control model selection, resource limits, notification delivery, and per-repository behavior.

Model Selection

Cloud agents use the default model configured for your workspace. You can override this at the workspace, repository, or individual run level.

Setting the Default Model

Go to Dashboard > Settings > Cloud Agents > Default Model. Select from any model available through the Creor Gateway or your configured BYOK providers.

Model	Speed	Quality	Cost	Best For
Claude Sonnet 4	Fast	High	$$	General-purpose. Best default for most tasks.
Claude Opus 4	Slow	Highest	$$$	Complex reasoning, large-scale refactoring.
Claude Haiku 3.5	Fastest	Good	$	Simple reviews, documentation, formatting.
GPT-4o	Fast	High	$$	Strong alternative to Sonnet for code generation.
GPT-4o mini	Fastest	Good	$	Budget-friendly option for simple tasks.

Per-Run Model Override

Override the model for a specific run by passing the model parameter in the launch request.

{

"repository": "github.com/acme/backend",

"branch": "main",

"prompt": "Refactor the payment processing module",

"model": "claude-opus-4-20250514"

}

Timeouts & Limits

Resource limits prevent runaway agents from consuming excessive credits or compute.

Setting	Default	Max (Pro)	Max (Enterprise)	Description
Run timeout	10 min	60 min	120 min	Maximum wall-clock time for a single run.
Max tokens	200,000	1,000,000	Unlimited	Maximum total tokens (input + output) per run.
Max tool calls	100	500	Unlimited	Maximum number of tool invocations per run.
Max file reads	50	200	Unlimited	Maximum number of files the agent can read.
Container CPU	2 vCPU	4 vCPU	8 vCPU	CPU allocation for the sandboxed container.
Container RAM	4 GB	8 GB	16 GB	Memory allocation for the sandboxed container.
Container disk	10 GB	20 GB	50 GB	Disk space for repository clone and temporary files.

Note

When a limit is reached, the agent stops gracefully and returns whatever results it has collected so far. The run status shows which limit was hit.

Monthly Spending Cap

Set a monthly spending cap to prevent unexpected charges. When the cap is reached, new agent runs are blocked until the next billing cycle or until you increase the cap.

# Set monthly spending cap via API

curl -X PUT https://api.creor.ai/v1/workspace/settings \

-H "Authorization: Bearer $CREOR_API_KEY" \

-H "Content-Type: application/json" \

-d '{

"cloudAgents": {

"monthlySpendingCap": 50.00

}

Webhooks

Configure webhooks to receive real-time notifications about cloud agent events. Webhooks are delivered as HTTP POST requests with a JSON payload.

Supported Events

Event	Trigger	Payload Includes
agent.started	Agent run begins execution	Run ID, repository, branch, model
agent.completed	Agent run finishes successfully	Run ID, artifacts, token usage, duration
agent.failed	Agent run fails or times out	Run ID, error message, partial results
agent.comment	Bugbot posts a PR comment	Run ID, comment body, file path, line number

Webhook Configuration

# Configure a webhook endpoint

curl -X POST https://api.creor.ai/v1/webhooks \

-H "Authorization: Bearer $CREOR_API_KEY" \

-H "Content-Type: application/json" \

-d '{

"url": "https://your-server.com/webhooks/creor",

"events": ["agent.completed", "agent.failed"],

"secret": "whsec_your_signing_secret"

Webhook payloads are signed using HMAC-SHA256 with the secret you provide. Verify the signature in the X-Creor-Signature header to confirm the request is from Creor.

# Verify webhook signature (Node.js example)

import crypto from 'crypto';

function verifyWebhook(payload: string, signature: string, secret: string): boolean {

const expected = crypto

.createHmac('sha256', secret)

.update(payload)

.digest('hex');

return crypto.timingSafeEqual(

Buffer.from(signature),

Buffer.from(expected)

);

}

Per-Repository Overrides

Override workspace-level settings for specific repositories. This is useful when different projects need different models, timeouts, or agent behaviors.

# Set per-repository overrides

curl -X PUT https://api.creor.ai/v1/repos/acme/backend/settings \

-H "Authorization: Bearer $CREOR_API_KEY" \

-H "Content-Type: application/json" \

-d '{

"cloudAgents": {

"defaultModel": "claude-opus-4-20250514",

"timeout": 30,

"maxTokens": 500000,

"instructions": "This is a Go backend. Use standard library where possible. Follow the patterns in CONTRIBUTING.md."

}

Repository-level settings override workspace-level settings. Run-level settings (passed in the launch request) override both.

Tip

Use the instructions field to give the agent project-specific context that applies to every run on that repository. This is especially useful for conventions, frameworks, and domain knowledge.

Environment Variables

Cloud agents can access environment variables you define in the workspace settings. These are injected into the container at runtime.

Variable	Purpose	Example
CREOR_AGENT_MODEL	Override default model from environment	claude-sonnet-4-20250514
CREOR_AGENT_TIMEOUT	Override default timeout (minutes)	30
Custom variables	Passed to the shell environment	DATABASE_URL, API_BASE_URL

Warning

Do not put sensitive credentials (database passwords, API keys) in cloud agent environment variables unless you trust the agent's sandboxed environment. The agent's shell can read all injected environment variables.

Full Settings Reference

Complete settings object with all cloud agent configuration options.

{

"cloudAgents": {

"defaultModel": "claude-sonnet-4-20250514",

"timeout": 10,

"maxTokens": 200000,

"maxToolCalls": 100,

"maxFileReads": 50,

"monthlySpendingCap": 100.00,

"instructions": "",

"webhooks": [

{

"url": "https://example.com/hooks/creor",

"events": ["agent.completed", "agent.failed"],

"secret": "whsec_..."

}

"env": {

"NODE_ENV": "production"

"repos": {

"acme/backend": {

"defaultModel": "claude-opus-4-20250514",

"timeout": 30,

"instructions": "Go backend. Follow CONTRIBUTING.md."

}