Continue an Agent Task

Send a follow-up prompt to an existing task, continuing the agent’s work with new instructions or additional context.

POST /api/tasks/{taskId}/continue

cURL

curl --location 'https://cloud.blackbox.ai/api/tasks/YOUR_TASK_ID/continue' \
--header 'Authorization: Bearer YOUR_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
    "currentPrompt": "Also add unit tests for the new payment module"
}'

Success Response

{
  "success": true,
  "message": "Task continue started successfully"
}

This endpoint allows you to continue an existing task by sending a follow-up prompt. The agent will pick up where it left off, applying your new instructions to the same repository and branch.

Authentication

To use this API, you need a BLACKBOX API Key. Follow these steps to get your API key:

1. Click on your Profile Image in the top right corner at cloud.blackbox.ai
2. Click on “BLACKBOX API Token” from the dropdown menu
3. Copy the existing token or click “Generate” if you don’t have one yet

Your API key will be in the format: bb_xxxxxxxxxxxxxxxxxxxxxx

GitHub Connection Required

For GitHub-related tasks: Make sure you have connected your GitHub account on https://cloud.blackbox.ai before continuing tasks that work with repositories. This is required for the agent to access and modify your repositories.

Headers

Authorization · string · required

API Key of the form Bearer <api_key>.

Example: Bearer bb_b41b647ffbfed27f61656049d3eaeef3d903cc503345d9eb80080d98bc0

Content-Type · string · required

Must be set to application/json.

Path Parameters

taskId · string · required

The unique identifier of the task to continue.

Example: ac2wbk3lXAufdSdIscsirg

Request Parameters

currentPrompt · string · required

The follow-up instruction or message for the agent. This is appended to the task’s history so the agent has full context of prior work.

Examples:

• "Also add unit tests for the new payment module"
• "The Stripe webhook handler is missing error handling, please fix it"
• "Summarize what was done so far"

selectedAgent · string

Override the agent for this follow-up. If omitted, the task’s original agent is used.

Available agents: claude, blackbox, codex, gemini, blackbox-grok

selectedModel · string

Override the model for this follow-up. If omitted, the task’s original model is used.

Example: blackboxai/blackbox-pro, blackboxai/anthropic/claude-sonnet-4.5

maxDuration · number

Maximum execution time in seconds for this follow-up. Defaults to the original task’s maxDuration.

keepAlive · boolean

Whether to keep the sandbox alive after this follow-up completes. Defaults to false.

installDependencies · boolean

Whether to run dependency installation before executing this follow-up. Defaults to false.

environmentVariables · string

Environment variables to inject into the sandbox for this follow-up, as a newline-separated KEY=VALUE string.

Example: "STRIPE_KEY=sk_test_xxx\nNODE_ENV=production"

globalInstructions · string

Additional system-level instructions to prepend to the agent’s context for this follow-up.

multiLaunch · boolean

Set to true to run this follow-up across multiple agents in parallel. Requires selectedAgents.

selectedAgents · array

Array of agent configurations for multi-agent follow-up mode. Required when multiLaunch is true.

Structure: Each object must contain:

• agent (string, required): The agent type
• model (string, required): The specific model to use

Minimum: 2 agents when multiLaunch is true

How Task Continuation Works

1. Follow-up stored: Your currentPrompt is appended to the task’s followupMessages array
2. Context rebuilt: The agent receives the full history of prior prompts and outputs as context
3. Sandbox resumed: A new sandbox is created (or reused) from the task’s existing branch
4. Agent executes: The agent applies your new instructions on top of prior work
5. Changes committed: Any new changes are committed to the same branch
6. Status updated: The task status returns to completed (or error) when done

cURL

curl --location 'https://cloud.blackbox.ai/api/tasks/YOUR_TASK_ID/continue' \
--header 'Authorization: Bearer YOUR_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
    "currentPrompt": "Also add unit tests for the new payment module"
}'

Success Response

{
  "success": true,
  "message": "Task continue started successfully"
}

Polling for Follow-up Completion

After calling continue, the task’s followupMessages array is updated with the new follow-up entry. Poll Get Task and inspect the last element of followupMessages:

Follow-up Message Structure

{
  "message": "Also add unit tests for the new payment module",
  "timestamp": "2025-11-21T10:35:00.000Z",
  "selectedAgent": "blackbox",
  "selectedModel": "blackboxai/blackbox-pro",
  "status": "completed",
  "statusMessage": null
}

status value	Meaning
processing	Agent is actively working on the follow-up
completed	Follow-up finished successfully
error	Follow-up failed; see statusMessage for details

Error Codes

Status Code	Error	Description
200	Success	Follow-up started successfully
400	Bad Request	Missing currentPrompt or invalid multi-agent configuration
401	Unauthorized	Invalid or missing API key
402	Payment Required	Insufficient credits for the selected model
404	Not Found	Task not found or access denied
500	Internal Server Error	File upload or sandbox creation failed

Related Endpoints

• Create an Agent Task — Start a new single-agent task
• Create a Multi-Agent Task — Start a new multi-agent task
• Get Task — Retrieve task details and follow-up statuses
• Stream Logs — Stream real-time execution logs
• Cancel Task — Stop a running task