Editions¶

TDB is an open-core product. The community edition is free, forever under AGPLv3. The paid editions add the connectors, auth, and compliance features you need to run TDB in production for a real team.

This page shows exactly what each edition includes. For pricing and a free 30-day evaluation, email us — we'll tailor a quote to your deployment and get you set up.

Tiers¶

Community — FreeSMBMid-MarketEnterprise

AGPLv3. Self-hosted. No credit card.

Built for developers who want to evaluate the concept with a single data source.

Data sources

CSV connector
1 registered source at a time

Auth

Single static API key

Query & API

REST query endpoint (SELECT only, 1,000-row cap)
Auto schema detection
OpenAPI / Swagger UI

MCP

query_source tool only
Lean Mode — smaller results for AI agents (rows as arrays, optional cell trimming), opt-in
Free VS Code extension — connect in one command; the same extension serves every edition

Audit

Local NDJSON audit log (append-only): every query, refusal and failure, with an option to keep values out

Install

Docker Compose one-liner
CLI (tdb serve, tdb register, tdb query)

Get started →

Commercial license. Self-hosted.

Everything you need to run TDB in production — multiple databases, governed MCP access, and a tamper-evident audit trail.

Data sources

PostgreSQL, MySQL, SQL Server, Snowflake, Parquet files, DuckDB databases
Unlimited simultaneous registered sources

Auth

JWT authentication
API key rotation and management
OAuth 2.1 with PKCE and dynamic client registration — MCP clients that support OAuth sign in through the browser instead of holding an API key
Per-key rate limiting
CORS configuration

Query & API

All Community features, plus:
Named views (YAML-defined, typed parameters)
PostgreSQL wire gateway — existing Postgres clients, BI tools and drivers connect to TDB unchanged apart from host, port and password, governed and audited

MCP

All 7 MCP tools: query_source, schema_source, preview_source, filter_source, aggregate_source, list_views, run_view (and run_approved where Mid-Market query approvals apply)
Schema annotations — describe what a cryptic column holds, so an AI assistant reads it correctly instead of guessing
Prompt injection filtering (input + output)
Per-key MCP tool allow-lists
Lean Mode — smaller results for AI agents, set per call, per API key or server-wide; schema detail levels for database-wide sources
Free VS Code extension — the same one community uses; your extra tools appear with no extension change

Access control

RBAC: read / readwrite / admin roles per API key

Audit

Signed hash-chained audit log (SHA-256, tamper-evident)
GET /v1/audit/verify — walk the chain, detect tampering
Configurable retention — seal the log into verifiable segments by size or age; TDB never deletes an entry

SIEM export

Splunk HEC audit export (POST /v1/audit/export)
S3 audit export — entries written to your own bucket, keyed by sequence range
Incremental by default: each destination keeps its own cursor, so a scheduled export ships only what it has not already sent
Datadog and OpenTelemetry are built on request — tell us which one you need

Observability

Prometheus metrics (GET /metrics)
Schema caching with configurable TTL
Health check endpoint

Contact us →

Commercial license. Self-hosted.

For regulated industries — healthcare-tech, fintech, and any team that needs a supported deployment and enterprise identity.

Everything in SMB, plus:

Support

SLA-backed support — next business day response
Onboarding call and deployment review

Enterprise identity

Single sign-on (OIDC), SCIM provisioning, and IdP groups in access rules
SAML (built on request)

Fine-grained access control

Column and row policies: per-key rules that withhold or mask columns (PII masking) and filter rows, on REST, MCP, views and the PostgreSQL wire gateway, with the rule recorded in the audit log (from 0.22.0)
Query approvals: a query that reads named columns waits for a person, and the audit log records the request, who decided and each run (from 0.32.0)
Tenants (on request; included in Enterprise)

Contact us →

Commercial license. Self-hosted.

For organisations that serve several teams, business units or customers from one deployment.

Everything in Mid-Market, plus:

Tenants

Tenants: one deployment, several teams or customers, each with its own sources, keys, views, policies, approvals and hash-chained audit log, chosen by the credential alone (from 0.33.0)
Your identity provider's groups place each person in a tenant
An operator role that manages tenants and reads none of their data, with its own audit log

Contact us →

Discounts¶

Annual subscriptions and larger / multi-seat deployments qualify for discounts. Email us to discuss terms for your team.

Edition comparison¶

The commercial editions ship the same product except for the Mid-Market features — column and row policies (from 0.22.0), single sign-on and SCIM (from 0.24.0) and query approvals (from 0.32.0, part of the policy file) — and tenants (from 0.33.0), which are Enterprise. A licence below the tier a feature needs refuses to start with it configured rather than serve without the rules. Otherwise SMB and Mid-Market differ on the support commitment, and Enterprise includes everything in Mid-Market.

Feature	Community	SMB	Mid-Market	Enterprise
Data sources
CSV connector	✅	✅	✅	✅
PostgreSQL	❌	✅	✅	✅
MySQL	❌	✅	✅	✅
SQL Server	❌	✅	✅	✅
Snowflake	❌	✅	✅	✅
Parquet files	❌	✅	✅	✅
DuckDB databases	❌	✅	✅	✅
BigQuery, JSON and NDJSON files, MongoDB	❌	On request	On request	On request
Simultaneous sources	1	Unlimited	Unlimited	Unlimited
Auth
Static API key	✅	✅	✅	✅
API key rotation & management	❌	✅	✅	✅
JWT authentication	❌	✅	✅	✅
OAuth 2.1 with PKCE	❌	✅	✅	✅
Single sign-on (OIDC) and SCIM	❌	❌	✅	✅
SAML	❌	❌	On request	On request
Per-key rate limiting	❌	✅	✅	✅
CORS configuration	❌	✅	✅	✅
Query & MCP
REST query endpoint (SELECT and CTEs)	✅	✅	✅	✅
PostgreSQL wire gateway (unchanged Postgres clients)	❌	✅	✅	✅
Row limit per response †	1,000	1,000 by default, configurable	1,000 by default, configurable	1,000 by default, configurable
Read past the row limit: page with a cursor, or stream as NDJSON or CSV	❌	✅	✅	✅
MCP tools	`query_source`	All 7	All 7, plus `run_approved` for callers an approval rule applies to	All 7, plus `run_approved` for callers an approval rule applies to
MCP tool allow-lists per key	❌	✅	✅	✅
Lean Mode: compact rows, cell trimming (per call / server default)	✅	✅	✅	✅
Lean Mode: per-key default, schema detail levels	❌	✅	✅	✅
Prompt injection filtering	❌	✅	✅	✅
YAML named views	❌	✅	✅	✅
Schema annotations for AI clients	❌	✅	✅	✅
Access control
RBAC (read / readwrite / admin)	❌	✅	✅	✅
Column and row policies, PII masking	❌	❌	✅	✅
Query approvals: a person approves reads of named columns	❌	❌	✅	✅
Suggested policies for personal-data columns	❌	✅	✅	✅
Tenants: several teams or customers on one deployment, each with its own audit log	❌	❌	On request	✅
Audit
Local NDJSON audit log	✅	✅	✅	✅
Hash-chained tamper-evident log	❌	✅	✅	✅
Log integrity verification	❌	✅	✅	✅
Audit log retention (sealed segments)	❌	✅	✅	✅
Splunk HEC export	❌	✅	✅	✅
S3 audit export	❌	✅	✅	✅
Incremental export cursor per destination	❌	✅	✅	✅
Datadog and OpenTelemetry export	❌	On request	On request	On request
Observability
Prometheus metrics	❌	✅	✅	✅
Schema caching	❌	✅	✅	✅
Health check	✅	✅	✅	✅
Support
Community (GitHub Issues)	✅	✅	✅	✅
SLA-backed (next business day)	❌	❌	✅	✅
Onboarding call	❌	❌	✅	✅

† Every edition caps a response at 1,000 rows by default — a deliberate safety default that keeps an AI agent from pulling an entire table. Community's cap is fixed; the commercial editions can raise or lower it with TDB_MAX_ROWS, and it applies to every path (REST, MCP tools, views). The commercial editions can also read further, from 0.31.0: page with a cursor, or stream a whole result as NDJSON or CSV, up to TDB_MAX_RESULT_ROWS.

Built on request¶

These are designed for but not built. Each one is built when a customer needs it, and we publish no dates for them:

Data sources: BigQuery; JSON and NDJSON files; MongoDB.
Audit export: Datadog and OpenTelemetry, beside the Splunk and S3 export that ship today.
Sign-in: SAML, beside the OIDC single sign-on that ships today.

If one of these decides your evaluation, tell us. A customer who needs one is what starts it.

The open-source guarantee¶

Features in the community edition are free forever under AGPLv3. We will never move a free feature behind a paywall. If you evaluate TDB today with CSV, those capabilities are yours permanently — no license change, no feature flag.

FAQ¶

Can I run the enterprise tier on-premises? Yes. TDB is self-hosted by design. The commercial license covers your on-premises deployment. There is no SaaS-only restriction.

How do evaluations work, and what happens when mine ends? Email us and we'll issue a 30-day evaluation licence at no charge — a dedicated, self-hosted image with the licence baked in, running entirely on your own infrastructure (no data reaches us). When it expires, TDB keeps running and /health stays healthy, but data and API requests return a clear license_expired response until you renew or convert. Need longer to evaluate? Just ask.

Do you offer annual or volume discounts? Yes — annual subscriptions and larger / multi-seat deployments qualify for discounts. Email us to discuss terms.

How much does it cost? Paid editions are a flat self-hosted license (no per-query or per-row metering). We share pricing by email so we can match the right edition to your deployment and include a free 30-day evaluation — get in touch.

Is the source code for the enterprise tier available? No. The enterprise tier is a closed-source commercial product. The community edition is fully open under AGPLv3.

I need BigQuery / Datadog / SAML now. When do those ship? They are built on request and not yet built. We deliberately don't publish dates we haven't committed engineering to — a date we miss is worse than no date, particularly in a product you're evaluating for compliance. If one of these is blocking your evaluation, email us: that's the signal that moves it up, and design partners get built first.