Works with the AI Agents you already use
Access Control Architect — Design Roles & Permissions
Access Control Architect — Design Roles, Permissions & Authorization: model least-privilege access and test policy decisions.
$9.99
Access Control Architect — Design Roles & Permissions
Example session with this skill installed
We are building a multi-tenant project management SaaS. Users can belong to multiple organizations. Owners invite members, assign roles, create projects, and share individual reports with external guests. Support staff can help customers but must not read customer data by default. Propose an authorization model, identify high-risk gaps, and give us a focused test plan.
- Read your context and instructions
- Compiled the access control architect
Recommended model
Use RBAC for stable organization-level capabilities and ReBAC for project/report relationships. Add ABAC only for trusted context such as support approval, resource sensitivity, or time-bounded elevation. Treat tenant membership as a verified server-side relationship.
Core policy map
| Subject | Action | Resource and scope | Decision rule |
|---|---|---|---|
| Organization owner | invite member, assign role | Own organization | Allow within the owner’s organization; block granting privileges above the owner’s delegation boundary. |
| Organization member | read project | Project in active organization | Allow only when current membership and project tenancy both match. |
| Project editor | update project | Assigned project | Allow for explicit editor relationship; reject client-supplied tenant or owner IDs as authority. |
| Report guest | read shared report | Specifically shared report | Allow read-only access while the share is active; deny sibling reports and administrative actions. |
| Support operator | access customer record | Customer tenant | Deny by default. Require approved, time-limited, scoped access with a reason and audit record. |
Enforcement and risks
Enforce every decision in the service/API layer, including object and export endpoints. UI hiding is not a control. Resolve tenant membership and resource relationships from trusted data. Keep role assignment separate from role definition and log grants, revocations, support elevation and sensitive reads.
Test cases
- A member cannot read a project in an organization they do not belong to.
- Changing a report ID cannot expose another tenant’s report.
- A guest can read only the report explicitly shared with them.
- An owner cannot grant a role above their delegation scope.
- Support access is denied without approval and expires at its configured time.
- Removing membership revokes access on the next authoritative check.
Assumptions to confirm
Define whether owners can share outside their organization, how quickly membership changes propagate, and who may approve support access. Validate the model against your actual data schema and enforcement code before rollout.
Connects securely to your tools. The creator never sees your data.
About this skill
Access Control Architect turns product requirements into explicit, least-privilege authorization designs. Use it to model roles and permissions, compare RBAC, ABAC, ReBAC and ACL approaches, define tenant and resource boundaries, review API or object-level checks, plan safe delegation, and build positive and negative policy tests.
The workflow separates authentication from authorization and treats server-side enforcement as authoritative. It makes assumptions visible when requirements are incomplete, maps subjects, actions, resources, relationships and context, then produces an implementation-ready policy model with enforcement points, migration steps, tests and governance guidance.
Use it for SaaS, APIs, internal tools, data platforms, AI agents and multi-tenant products. It works from the requirements, diagrams, code excerpts or configuration you provide; it does not claim live access to repositories, IAM consoles, logs or cloud environments.
This skill provides design and review guidance based on the context and artifacts supplied in the current session. It does not independently connect to repositories, identity providers, policy engines, cloud consoles, production systems, or audit logs. Recommendations must be validated against the actual implementation, data model, threat model, and applicable requirements. It does not certify compliance or guarantee that a system is secure.
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 1 month ago
Permissions
File access and shell use are only needed when the user supplies a repository or asks to run tests. No external service account, network connection, or environment-variable access is required by default.