- Home
- Skills
- Data & Databases
- 🛡️ Supabase RLS Doctor
More screenshots
Works with the AI Agents you already use
🛡️ Supabase RLS Doctor
The scanner flags four kinds of review markers: service-role references, broad true predicates, grants, and policy/storage statements.
$19
🛡️ Supabase RLS Doctor
Example session with this skill installed
Review the bundled synthetic fixtures/risky-sample.sql. Separate scanner markers from confirmed deployed risks; do not connect to Supabase or run SQL.
- Read your context and instructions
- Compiled the supabase rls doctor
Synthetic fixture walkthrough
This is an illustrative agent review grounded in the bundled fixture, not a customer production audit or proof of deployed exposure.
| Marker | Fixture line | What needs review |
|---|---|---|
| RLS001: service-role reference | 4 | A comment names the service role. This does not establish a leaked credential. |
| RLS002: broad true predicate | 3 | Check whether public reads are intentional for this table and role. |
| RLS003: grant | 2 | Review the broad grant alongside the complete schema and policies. |
| RLS004: policy/storage marker | 3 | Inspect the policy in context; a policy declaration is not itself a vulnerability. |
Evidence boundaries
Only the synthetic SQL file was inspected. No database connection, SQL execution, credentials or deployed-role tests were used. The fixture does not establish the state of a real project.
Next steps
Supply the complete relevant migrations and intended role/tenant rules. Review policy coverage and ownership manually; distinguish absent evidence from confirmed missing RLS. Propose changes only after that review and request approval before applying them. No automatic repair was made.
Connects securely to your tools. The creator never sees your data.
About this skill
Review Supabase RLS evidence before proposing changes
Inspect supplied SQL migrations, policy files and client source with a local Python helper and guided checklist. The scanner flags four kinds of review markers: service-role references, broad true predicates, grants, and policy/storage statements.
Separate markers from confirmed risk
A marker is not proof of a leaked key or an exposed table. Missing RLS, tenant isolation, storage ownership and deployment drift require contextual manual evidence. The agent adds impact, proposed remediation and verification steps only where the supplied evidence supports them; unavailable deployed state remains unverified.
Local, read-only workflow
The scanner uses the Python standard library. No credentials, paid APIs, database connection, SQL execution or automatic policy changes are required. Use an agent with local file access and Python execution; without Python, follow the checklist and report that automated scanning was not performed.
Current-delivery caution
The original v1.0.0 scanner can include surrounding source text in reports and does not reliably distinguish incomplete input coverage. Review those reports locally before sharing, and do not use that version as an unattended security or coverage gate. A corrective maintenance package is being submitted; delivery verification is still pending.
What you receive
SKILL.md instructions, the Python scanner, two synthetic fixtures, tests, a manual audit checklist and remediation templates. The workflow is not security certification and cannot establish the state of a live Supabase project from local source alone.
- Four regex rules identify review markers, not confirmed vulnerabilities or proof of safety.
- Missing RLS, deployed exposure, tenant isolation, bucket access and migration drift require manual evidence.
- No database connection or SQL execution.
- Current v1.0.0 can include source excerpts and does not reliably report incomplete coverage; inspect reports before sharing. Corrective package delivery is pending.
- Agent loading compatibility is not proof of execution in every host.
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 today
Requires
Permissions
File Scopes
Read-only inspection first. The bundled scanner reads matching files and prints markdown or JSON findings. It installs nothing, transmits nothing, and modifies nothing. Any write, install, deploy, payment, delete, reset, live-account action, or external network lookup requires explicit user confirmation.