v0.9.0 · 28 May 2026

In short

• We launched API v1 for chats, sources and documents, so TasK can be connected to your services.
• We opened the public projects we use in TasK development: task orchestration, coding standard, todo-md and git-workflow.
• The new AI-agent team page shows the AI-agent roles that help develop TasK.
• On the AI-agent team page, we’ve prepared a chat with the AI-agent team about the product’s capabilities. We’re still testing it; once we’re sure it works as expected and is safe for us 😎, we’ll switch it on.
• We added TasK use cases to the Help Center — so it is easier to choose a starting task and understand what materials to upload to a project.
• At users’ request, we raised the maximum upload file size from 100 to 500 MB.

What’s new in the API

API v1 helps connect TasK to your services and automations: you can work with projects, chats, sources, documents and webhooks within the user’s access rights.

• Projects — GET /v1/projects, POST /v1/projects, GET /v1/projects/{uuid}, PUT /v1/projects/{uuid}, DELETE /v1/projects/{uuid} — create and update the projects that store sources and chats.
• Chats — GET /v1/chats, POST /v1/chats, GET /v1/chats/{chatUuid}, PATCH /v1/chats/{chatUuid}, DELETE /v1/chats/{chatUuid} — create chats inside a project and manage them.
• Chat messages — POST /v1/chats/{chatUuid}/messages and GET /v1/chats/{chatUuid}/messages — send requests and browse message history.
• Sources — POST /v1/projects/{projectUuid}/source-files, POST /v1/projects/{projectUuid}/source-urls, POST /v1/projects/{projectUuid}/source-contents, GET /v1/projects/{projectUuid}/sources, GET /v1/sources/{uuid} — upload files, URLs and text, then track processing status.
• Documents — GET /v1/projects/{projectUuid}/sources/{sourceUuid}/documents, GET /v1/projects/{projectUuid}/documents/{uuid}, GET /v1/projects/{projectUuid}/documents/{uuid}/content — list Markdown documents generated from a source, read their metadata and fetch content: these are ready text representations of sources that TasK uses for RAG and LLM answers.
• Fragment search — POST /v1/projects/{projectUuid}/chunks/search — find relevant document fragments: the query is embedded and matched against prepared chunks by vector similarity. These fragments can be passed to the LLM as additional context without sending the whole project context.
• Webhooks — POST /v1/webhooks, GET /v1/webhooks, DELETE /v1/webhooks/{uuid} — receive source processing status notifications and trigger the next steps in your services without polling the API.

Create an API key in the web interface: open Dashboard → API keys, set a name, expiration and scopes. The token is shown only once — save it and include it in requests.

AI-agent team page

On the AI-agent team page, we show the AI-agent roles that help build TasK. It is a public map of the roles: who is responsible for what, how the roles are configured and how they differ from one another.

• A live example of AI-first development — see which areas of work we are trying to delegate to AI: product, marketing, design, development, testing, review and support.
• Parameters and personal profiles — each role includes a description of its responsibilities, working style, behavior profile and personality profile.
• An open experiment — we are testing how behavior and personality settings affect decisions, copy, reviews and collaboration between AI agents during development.

We prepared a chat with the AI-agent team right on this page: it will let you ask about TasK capabilities. We are still testing the answers and checking that it is safe enough for us. Once we are sure it works as expected, we will enable it.

Use cases in the Help Center

We added use cases to show which TasK tasks you can start with and what materials to upload.

• Project knowledge base — bring recordings, documents, links and notes into one project and ask questions across the materials.
• Long-form video insights — get the key points from a long recording, find useful fragments and decide what to watch in detail.
• Multilingual knowledge base — ask in your own language while the source materials are in another language.
• Expert opinion synthesis — compare interviews, articles and videos to see common ground, disagreements and practical takeaways.
• Knowledge base API for AI agents — connect TasK knowledge to apps and AI agents through the API.

Public projects for development

We moved some TasK development tools into separate public repositories. They help keep tasks, conventions and AI-agent work more predictable.

Read more in the article about how we use AI agents in development.