Task Agent
About Task Agent
Plan, execute, and deliver complete tasks with tools, files, and lasting context.
Hast Agent is an AI workspace for complete tasks. These docs call its task interface Task Agent. It organizes plans, tool calls, execution, and deliverable files around a goal, especially for work with multiple steps and revisions based on results.
Describe your goal, review the plan, then confirm execution. You can follow up, revise requirements, or download generated files.
Suitable tasks
| Scenario | Work to delegate | What to check |
|---|---|---|
| Market and competitor research | Collect sources, compare options, prepare briefs | Sources, dates, and evidence |
| Data analysis | Organize tables, compare metrics, explain findings | Definitions, calculations, outliers |
| Content and documents | Summarize, draft reports, revise from feedback | Completeness, audience, format |
| Development and operations | Work with code, browsers, or recurring tasks within available permissions | Actual results and required human actions |
Specify sources, scope, and output format. For example: “Compare three options using these materials, produce a pros-and-cons table, and list questions still requiring verification.”
Continue in the same workspace
Replies and files remain in the workspace. Inspect execution, request additions or revisions in the original conversation, and download final outputs. See files.
Agents can use code, web search, browsers, and connected apps. External apps require authorization; the workspace shows the available connectors. See apps and capabilities.
Reuse context and schedule recurring work
Save project context, business rules, and team preferences in memory to reduce repetition. Review and maintain that memory; add temporary requirements in the current conversation.
Use schedules for daily reports or recurring summaries. Set instructions, frequency, and timezone, then check each run and refine the instructions based on output.
Complete a task
- Start a conversation with requirements and optional files.
- Review the first plan and fill information gaps.
- Select the action to start implementation when ready.
- Watch execution and inspect replies and files.
- Request revisions in the same conversation, or start a new one for different work.
Follow the quickstart for a document task, or use schedules for recurring work.
Programmatic access
The public API uses personal API keys to submit messages, read events, and manage files. API-created sessions are listed separately; the API list does not combine all historical web conversations.
Get started
Choose a task with complete input and easily checked output, then follow your first task. For integration, read the API quickstart. See plans and usage for account information and the Hast Agent product page for more scenarios.