Delega

Delega is a tool for AI agents that need to keep working state between sessions. It gives Claude Code, Codex, Cursor, and custom agents one shared task list, durable context, and decision history so work can resume instead of restarting. The product also frames its use case as solving the problem of agents starting from zero and repeatedly rediscovering what happened before.

Its core functions center on task context and handoffs. The page says agents can recover the last decision, files, and next step before touching code, and that work can move between agents without using the message bus. It mentions named operations such as list_tasks, get_task_context, and update_task_context, along with a durable context record for each task and an audit trail for future sessions. Other described features include delegation chains that keep parent and child tasks, assignees, and context linked, plus a browser task console for filtering, inspecting, recovering, and deep-linking agent work. The console also supports actions such as create, complete/reopen, reassign, and release, and it shows context history, comments, subtasks, and chain information.

Delega is available through MCP, CLI, API, and Python SDK, and it uses one hosted memory layer across those interfaces. The page includes examples for Python, TypeScript, cURL, CLI, and MCP, and it states that the browser console is available at /tasks. It also says portable memory is shared across REST, MCP, CLI, Python SDK, the browser console, export API, and JSONL sync. Pricing is described as 1,000 free tasks per month, with self-serve access available without a card, and the site also offers founder-led setup help for AI dev teams.