# agent-taskbay

> A durable, human-in-the-loop console for A2A agents. Review approvals, follow long-running tasks, and preserve conversation history across agent sessions.

Full documentation: https://taskbay.allsrc.dev/llms-full.txt

- [Documentation](https://taskbay.allsrc.dev/docs): Find the page that answers your question. Each page answers one. Start with your role, or scan the task list below.
- [Quickstart](https://taskbay.allsrc.dev/docs/quickstart): someone who wants to see Agent Taskbay work on their own machine. At the end you will have run the console, connected a sample A2A agent, sent it a task, approved an action as a reviewer, and seen the result recorded.
- Operator handbook
  - [Operator handbook](https://taskbay.allsrc.dev/docs/handbook): Configure Taskbay, secure human access, connect protected agents, and understand the supported feature set.
  - [Environment map](https://taskbay.allsrc.dev/docs/handbook/environment-map): Find every Taskbay runtime variable, understand secret boundaries, and apply configuration consistently.
  - [Secure Taskbay with OIDC](https://taskbay.allsrc.dev/docs/handbook/secure-oidc): Configure HTTPS, authorization code with PKCE, server-provisioned membership, roles and scoped grants.
  - [Connect an OAuth-protected agent](https://taskbay.allsrc.dev/docs/handbook/oauth-agents): Register a protected Agent Card, configure a client-credentials binding, verify discovery, and manage rotation.
  - [Capability map](https://taskbay.allsrc.dev/docs/handbook/capabilities): Every major Taskbay capability, its setup path, and the boundaries of current support.
- Concepts
  - [Overview: what Agent Taskbay is](https://taskbay.allsrc.dev/docs/concepts/overview): For readers who want the mental model before they install anything: what the project does, what it leaves to other systems, and how the pieces fit together.
  - [Durable tasks](https://taskbay.allsrc.dev/docs/concepts/durable-tasks): For readers who want to know why a task survives a closed browser, what the console does when something goes wrong, and what "uncertain" means.
  - [Identity and access](https://taskbay.allsrc.dev/docs/concepts/identity-and-access): For readers who need to understand the three different meanings of "authentication" in Taskbay, and how organizations, roles, teams and grants decide what a person can see and do.
  - [Human in the loop](https://taskbay.allsrc.dev/docs/concepts/human-in-the-loop): For readers who want to understand how work reaches a person, how an approval is made safe, and how ownership, escalation, audit and notifications fit together.
  - [Rich content](https://taskbay.allsrc.dev/docs/concepts/rich-content): For readers who want to know how Taskbay shows what agents send (text, data, files, forms, generated interfaces), and the firm rule behind all of it: agent content is data, never code.
  - [Design decisions](https://taskbay.allsrc.dev/docs/concepts/design-decisions): For readers evaluating Taskbay who want to know why it is built this way, and what each choice costs. Each row links to the longer archived decision record.
- Guides
  - [Use the console: chat, tasks, flows, audit and settings](https://taskbay.allsrc.dev/docs/guides/use-the-console): anyone who works in the Taskbay console day to day. You will have a map of every screen, know what each control does, and know which actions your role allows.
  - [Connect an agent](https://taskbay.allsrc.dev/docs/guides/connect-an-agent): an administrator adding an A2A agent to Agent Taskbay. At the end the agent is registered, reachable, authenticated if it needs to be, and visible to the people who should be able to use it.
  - [Build an agent for Taskbay](https://taskbay.allsrc.dev/docs/guides/build-an-agent-for-taskbay): an author of an A2A agent who wants people to answer it through Agent Taskbay: with a typed form, an A2UI surface, or an approval step. At the end you know which extensions to advertise, what to send, what comes back, and how the reference fixture does it.
  - [Approvals and ownership](https://taskbay.allsrc.dev/docs/guides/approvals-and-ownership): operators, reviewers and administrators who use the console day to day. At the end you can find work that needs you, decide an approval, take ownership of a task, set a due time, and configure where overdue work goes.
  - [Notifications](https://taskbay.allsrc.dev/docs/guides/notifications): people who use the console (to know what they will be told) and administrators who want the same notifications in a team chat or another system through a signed webhook. At the end you have the in-app inbox understood and, optionally, a verified webhook receiver.
  - [Sign in and roles](https://taskbay.allsrc.dev/docs/guides/sign-in-and-roles): For operators who need people to sign in to Agent Taskbay. At the end you will have OIDC sign-in working, a first administrator provisioned, and teams and grants set up.
  - [Give people access to agents and skills](https://taskbay.allsrc.dev/docs/guides/teams-and-grants): administrators who need operators and viewers to see and use specific agents. You will have a team, its members, and a grant that lets the team work with one skill of one agent, plus a way to check and revoke it.
  - [Agent credentials and outbound policy](https://taskbay.allsrc.dev/docs/guides/agent-credentials): For operators whose agents require authentication, or who run Taskbay outside a trusted laptop. At the end you will know where credentials are stored, how to add, rotate and revoke them, and which outbound network rules apply.
  - [Receive push callbacks from agents](https://taskbay.allsrc.dev/docs/guides/agent-push-callbacks): operators who want agents to report task changes to Taskbay as they happen, instead of waiting for the next poll. You will have A2A push notifications turned on, a callback URL that agents can reach, and a way to tell whether callbacks arrive.
  - [Configuration recipes](https://taskbay.allsrc.dev/docs/guides/configuration-recipes): operators and developers who want a working set of A2A_ variables for a common setup without reading every row of the reference. You will have a copy-pasteable environment for your case, the commands to start it, and a check that tells you it is working.
  - [Production deployment](https://taskbay.allsrc.dev/docs/guides/production-deployment): For operators putting Agent Taskbay on a shared server. At the end you will have a web process, a worker process and a PostgreSQL database running behind HTTPS, with people signing in through OIDC.
  - [Upgrading and backups](https://taskbay.allsrc.dev/docs/guides/upgrading-and-backups): For operators of a running Agent Taskbay. At the end you will know what to back up, how to upgrade without losing data, and how to restore.
  - [Try the AG-UI adapter](https://taskbay.allsrc.dev/docs/guides/enable-ag-ui): a developer who wants an AG-UI client to drive an A2A agent that is already connected to Agent Taskbay. You will have the adapter turned on, a first run streamed from a sample agent, and an agent question answered through an interrupt.
- Reference
  - [Configuration reference](https://taskbay.allsrc.dev/docs/reference/configuration): For operators who need to know exactly what each A2A_ environment variable does, its default and what happens when it is wrong. Every entry
  - [CLI reference](https://taskbay.allsrc.dev/docs/reference/cli): For anyone who needs the exact commands, flags and exit behavior: the agent-taskbay launcher, every npm run script, and the operational
  - [HTTP API reference](https://taskbay.allsrc.dev/docs/reference/http-api): For people who script against Agent Taskbay or build on it: every route under /api, with its method, who may call it, what it accepts and what it
  - Extensions
    - [Extensions reference](https://taskbay.allsrc.dev/docs/reference/extensions): For agent authors and operators: the A2A extensions Agent Taskbay defines or understands, how an agent opts in, and what happens when it does not.
    - [Structured form extension (v1)](https://taskbay.allsrc.dev/docs/reference/extensions/structured-form): For agent authors: how to ask a person for typed input (or offer a start-of-task form) so the console renders real fields instead of a free-text box.
    - [Approval request extension (v1)](https://taskbay.allsrc.dev/docs/reference/extensions/approval-request): For agent authors: how to ask a human reviewer to approve one exact action, and what you can rely on when they do. The agent proposes; it has no authority to decide.
    - [Skill routing extension (v1)](https://taskbay.allsrc.dev/docs/reference/extensions/skill-routing): For agent authors and administrators: how Agent Taskbay can restrict a person to one skill of an agent, and why that only works if the agent enforces it.
  - [AG-UI adapter reference](https://taskbay.allsrc.dev/docs/reference/ag-ui): For developers who want an AG-UI client to drive an A2A agent through Agent Taskbay: how to enable the adapter, the request it accepts, how task state maps to AG-UI events, and
  - [A2UI reference](https://taskbay.allsrc.dev/docs/reference/a2ui): For agent authors: which parts of the A2UI interface format Agent Taskbay renders, what it refuses, and the safety rules that hold whatever an agent sends.
  - [Data model reference](https://taskbay.allsrc.dev/docs/reference/data-model): For operators, contributors and integrators who need to know what Agent Taskbay stores, how records are identified, and which rules the database itself enforces.
  - [Compatibility and known limitations](https://taskbay.allsrc.dev/docs/reference/compatibility): For anyone deciding whether Agent Taskbay fits their environment and agents: supported runtime versions, which A2A versions, transports and authentication schemes the gateway
  - [Glossary](https://taskbay.allsrc.dev/docs/reference/glossary): anyone reading the Taskbay docs who meets an unfamiliar term. You will have a short definition of each term and a link to the page that explains it in full.
- Operations
  - [Running and workers](https://taskbay.allsrc.dev/docs/operations/running-and-workers): For operators who need to know what Agent Taskbay's processes do, how to run the worker separately, and how to tell whether it is healthy.
  - [Troubleshooting](https://taskbay.allsrc.dev/docs/operations/troubleshooting): For anyone staring at an error from Agent Taskbay. Find the message or symptom, learn the cause, apply the fix.
- Security
  - [Threat model](https://taskbay.allsrc.dev/docs/security/threat-model): operators and security reviewers deciding whether Agent Taskbay's controls fit their deployment. You will have the assets, trust boundaries, threats, the control for each, and the residual risk.
  - [How agent credentials are protected](https://taskbay.allsrc.dev/docs/security/service-identity): a security reviewer or operator who wants to know how Taskbay stores and uses the credentials it holds for your agents. You will have the storage model, the network rules, and what is not covered. For the setup steps, use [agent credentials](../guides/agent-credentials.md).
- Project
  - [Status and roadmap](https://taskbay.allsrc.dev/docs/project/status-and-roadmap): someone deciding whether Agent Taskbay is mature enough for their use. You will have what works today and how to verify it, what is partial, what is planned, and what is parked.
- Contributing
  - [Development setup](https://taskbay.allsrc.dev/docs/contributing/development): someone who wants to change Agent Taskbay. You will have it running from source, know which test layer to use, and know the code boundaries your change must respect.
  - [Writing and updating the docs](https://taskbay.allsrc.dev/docs/contributing/documentation): anyone changing behavior or writing a docs page. You will know where a change belongs, which page to update, and the style rules.