Skip to content
TanomisDocs
Open Console

Tanomis documentation

Run AI agents as a governed service.

You write the agent in Python. Tanomis packages it, runs it for each of your customers in its own sandbox, and gives every action an identity, a policy check, and a record. Warm starts, approvals, and audit come with the platform, not with your code.

Deploy your first agentabout 10 minutesTour the Console
Four commands, one running agent
tanomis login
tanomis agent new support-copilot --harness claude-agents-sdk
tanomis agent deploy ./agents/support-copilot --tenant northwind-sandbox
tanomis session run support-copilot --prompt "Draft a reply to ticket 4821."
  • Sandbox per session
  • Warm start under a second
  • Signed, immutable builds
  • Tamper-evident audit ledger
  • Your tenants, your policy

How it works

Section titled “How it works”

Your source becomes an immutable version. A tenant is enabled for it. Your application opens a session and starts runs. Every run leaves evidence. Nothing skips a step, and nothing routes traffic until you enable it.

  1. SourceYour Python agent, in your repository.
  2. VersionAn immutable build plus the contract you declared.
  3. EnablementWhich version each tenant serves, canary or baseline.
  4. SessionA conversation or unit of work, warm-started in under a second.
  5. RunOne request, streamed, pausable for a person.
  6. EvidenceWho asked, what was allowed, what happened, signed.

What teams build on it

Section titled “What teams build on it”

Three agents that use most of the platform. Each one starts from a guide you can follow today.

Customer-facing

Support copilot

Reads the ticket and your knowledge base through a governed connector, drafts the reply, and pauses for a person before issuing a refund.

  • MCP connector
  • Approval on refund
  • Streaming
Build it
Back office

Invoice reconciler

Picks up a unit of durable work, matches it against the ledger over several sessions, and settles it once. No person ends it by hand.

  • Durable work
  • Session files
  • Egress allowlist
Build it
Engineering

Code review assistant

Reads a pull request through the GitHub connector, hands the security pass to a second agent under a narrower grant, and posts one review.

  • Delegation
  • Schema pins
  • Canary rollout
Build it

Everything the platform gives you

Section titled “Everything the platform gives you”

The feature set, in the order you meet it.

Buildyour code, your model
Harnesses and modelsClaude Agent SDK or Deep Agents; any model on the tenant’s allowlist, routed through one gateway.
Governed toolsMCP connectors with an explicit allowlist. Agents never launch servers themselves.
One manifestModels, tools, hosts, limits, and approvals declared once in agent.yaml.
Shipimmutable, per tenant
Versions and canaryOne command builds, registers, snapshots. Roll out by tenant share; roll back by re-enabling.
Warm startsA golden snapshot restores the harness in under a second, with no tenant data inside.
CI and automationDry-run on every pull request; deploy with an organization-owned credential.
Runfrom your product
Sessions and streamingHTTP in, events out. Reconnect and replay safely; attach files with integrity checks.
Approvals and questionsThe run pauses, survives disconnects, and resumes on the first answer, from the Console or your app.
DelegationOne agent hands work to another under a live grant that only narrows.
Trustwhat auditors ask for
Policy ceilingsNetwork egress, model, and tool ceilings per tenant; the manifest can only ask for less.
EvaluationTrace every run, score a candidate against the baseline, gate promotion on the verdict.
Audit ledgerAppend-only, hash-chained, signed per tenant. Export an auditor package in one step.

Every run is governed

Section titled “Every run is governed”

Whatever the agent does, the same six things happen around it. The example below follows one run end to end: a support copilot answering a ticket for the customer bluewater-bank, that needed a refund approved before it could finish.

Example: one governed runA support copilot answering a ticket for one customer.
  1. Identity
    Support team memberThe authenticated principal starts the request.
  2. Context
    Northwind / bluewater-bankYour organization and the customer tenant the run serves.
  3. Policy checkpoint
    Action allowedPrincipal authority, tenant policy, and the version contract intersect.
  4. Agent action
    support-copilot@0.4.0The exact version that acted, attested by the platform.
  5. Human decisionwhen required
    Refund approvedWhen supported and required, a durable interaction pauses and resumes the run.
  6. Evidence
    succeededA run_completed event can signal completion; the run resource status is authoritative.

Read the whole model in Policy, approvals, and evidence.

Start where you are

Section titled “Start where you are”
Integrating an applicationInvoke from your applicationThen Sessions and runs and the Platform API.
Administering a workspaceTour the ConsoleThen Members and access and Audit and evidence.

Working with a coding agent such as Claude Code, Codex, or Cursor? One command installs skills that teach it to operate Tanomis and registers a documentation MCP server: see Build with a coding agent.

Reference

Section titled “Reference”
CLIEvery tanomis command and flag.agent.yamlManifest fields and defaults.Platform APIEndpoints, headers, errors.EventsRun event types and replay.