- Home
- Skills
- Data & Databases
- Supabase Tenant Isolation Test Builder
Works with the AI Agents you already use
Supabase Tenant Isolation Test Builder
Build explicit CRUD access matrices and pgTAP regression tests for a local Supabase-style database. Cover allowed access as well as cross-tenant denials.
$5
Supabase Tenant Isolation Test Builder
Example session with this skill installed
Synthetic example: use the included membership/shared-document contract to generate tests in a new output directory. Preserve legitimate own-tenant and shared access, test foreign-record denials and revocation, inspect the SQL and do not execute it against a database.
- Read your context and instructions
- Compiled the supabase tenant isolation
Synthetic generation-only example: isolation.test.sql with a 69-assertion plan, MATRIX.md and manifest.json with executed: false and input/SQL hashes. No database has been tested by generation. A later SQL test needs separate approval and complete TAP results.
Connects securely to your tools. The creator never sees your data.
About this skill
Changing tenant policies? Test what users can actually read and change.
For developers building multi-tenant apps with Supabase-style PostgreSQL authorization. Use your coding agent to turn agreed access rules into a reviewable test matrix and reproducible SQL tests—not just a checklist saying that RLS is enabled.
What you get
- An agent workflow to confirm business rules, actors, membership and intentional sharing before testing.
- A dependency-free Node.js generator for SELECT, INSERT, UPDATE and DELETE cases.
- A readable matrix, pgTAP SQL and a manifest with input/SQL hashes. Generation is not execution.
- Exact expected rows or permission errors, runtime role/nonprivileged-context checks and per-case rollback.
- A clearly synthetic membership/shared-document example and a configuration reference.
A practical workflow
Declare two tenants, their legitimate users and an unrelated user. Include allowed own-tenant operations, foreign-record attempts and intentional shared access. Add tenant reassignment and membership revocation cases where they belong in your contract.
The agent helps map your development schema and synthetic fixtures, then runs the supplied generator. You inspect the matrix and SQL. Database execution is a separate action requiring your permission. Failures are reported against the declared contract with reproduction evidence; policies and grants are not silently repaired.
Unlike an open-ended prompt, the package supplies a bounded generator with validated expectations and repeatable output. It does not infer the right business policy. Missing rules remain unresolved; an expected zero rows is not an unknown result.
Before you use it
Generation requires Node.js 22+ and an agent able to read files and run a temporary script. SQL execution additionally requires an approved disposable/local development database, pgTAP in extensions, Supabase-style roles/Auth helpers, reviewed fixtures and a TAP harness. It is not a hosted database service or a one-click project installer.
Evidence and boundaries
Version 0.1.1 generated SQL ran unchanged in one author-operated replay on pinned Basejump migrations: 266 passing assertions on the correct control and seven detected deliberately seeded in-scope regressions. Fixtures were synthetic and Auth helpers were SQL test doubles. This is not independent customer validation, a discovered upstream vulnerability or a full Supabase/API test.
Version 1 supports bounded structured CRUD, scalar equality filters and declared projections. Expected SQLSTATEs are 00000 and 42501; custom errors such as P0001 need separate tests. Trigger/external effects are not made universally reversible by transactions.
This is a regression-test builder, not an automatic security audit, JWT verifier or production repair tool. It does not cover arbitrary RPC/views/functions, Storage, Realtime or every application workflow. Cross-agent compatibility, actual Agensi delivery, time savings and customer demand remain unverified. Supabase and pgTAP already provide free testing tools; this package adds a focused contract-to-matrix workflow.
Matrix v1: at most 100 cases, 16 actors and eight targets; scalar equality filters, explicit projections and expected SQLSTATE 00000/42501 only. P0001 needs separate custom tests. No automatic rule discovery, full security audit, hosted/production testing, guaranteed rollback of external effects or cross-agent/marketplace compatibility.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 21 days ago
Requires
Permissions
Allowed Hosts
File Scopes
The generator has no database or network connection and does not modify policies. Agensi provides skill material; model-provider data handling applies separately. Database execution, dependency installation and policy changes require separate authorization. No production credentials or input upload are required.