Help Center

Tangigo is a platform for building software with AI agents that answer to your organisation. It holds an org-scoped registry of what your organisation has agreed — architecture principles, APIs and contracts, data models, capabilities, components, integrations, skills and policies — and a control plane that decides who, person or agent, may do what. Agents are grounded in that registry, and the work they do ends in a pull request on your own repository.

Sign up at /auth/signup with an email address or a Google, GitHub or Microsoft account. The Free tier needs no credit card; you create your organisation in the onboarding wizard and can start a product straight away.

They are the three levels of the model. An organisation owns the registry and the control plane and holds your members. A product is a thing you are building. A module is a deliverable piece of that product with its own repository, environments, backlog, tests and documents. Today the platform is aimed at software products — web, mobile, backend and similar.

It is the record of what your organisation has: agent definitions, skills and tools, APIs and contracts, architecture principles and standards, capabilities, data models, reusable components, integrations, policies, compliance frameworks and credential bindings. Every write creates a new version, each entry has an owner, a lifecycle from proposed through approved, active and deprecated to retired, and a list of what uses it. It is also what agents are grounded in, so answers come from the registry rather than from a chat thread.

An agent definition bundles skills, a control-plane profile, a default autonomy level and a system prompt. You provision it into named agent instances that are principals in their own right, with their own identity in the audit trail. A squad is a mix of org members and agent instances in roles — architect, developer, tester, BA, UX — staffed onto a backlog item. Agents attach skills, never raw tools.

There is exactly one build path. The story is dispatched to a queue and picked up by an isolated job that clones the module repository, runs the architect, developer and tester over a single working tree, commits, pushes and opens one pull request. Progress is checkpointed, so a build survives you closing the browser, and it streams live into the squad room. After the pull request merges, a verification pipeline runs the test matrix against a baseline held per module and environment, and reports honestly — a test type that timed out or was skipped does not count as verified.

When work deviates from one of your organisation principles, a checkpoint card appears inline in the flow — in the wizard, in grooming or in the squad room. It states the rule, the deviation and who must approve. The outcome is a rejection, a scoped and time-boxed dispensation, or a relaxation of the principle itself. Either way an ADR is written and the registry is updated, so the decision is recorded rather than remembered.

Specs, stories, ADRs, architecture views and policies are Markdown documents edited in the product, with typed references between them, versions and an approval workflow. Diagrams are draw.io, versioned and referenced by id from documents, and ERDs are generated from the data models in your registry. One scope-filtered search runs across documents and diagrams, and it is the same index agents are grounded in.

Yes. An MCP server is registered as a skill source: its tools are discovered and installed as skills your agents can attach. A published API entry in the registry generates a skill and its HTTP tools in the same way. Credentials are resolved per call from a credential binding.

Secrets are write-only. A credential binding in the registry holds an opaque reference and metadata that your members can read; the secret itself is never returned and is resolved only at the moment a call needs it. Access to bindings is governed by the control plane and recorded in the audit trail.

There are four tiers — Free, Pro, Team and Enterprise. The Free tier needs no credit card. See /pricing for what each tier includes.

Open a request at /support, or email support@tangigo.com. Either way it becomes a ticket our team reads. If you already have an account, you can also ask the assistant inside the product.