Skip to content

mcp_authorization v0.7.1

Cookbook Reference Releases GitHub

1. Expose your first tool (read-only, end to end)
2. Hide an entire tool behind a permission
3. Hide a tool behind an account feature, not a role
4. Hide one input field from some users
5. Give privileged users a richer output shape
6. Pre-fill a field from the current user
7. Validate input with constraints
8. Share a type across handlers
9. Model success-or-error as a discriminated union
10. Require a field only when another is present
11. Change the tool description by role
12. Serve different tool surfaces from one app (domains)
13. Invent your own predicate vocabulary
14. See exactly what a role sees (debugging)
15. Mark a tool read-only / destructive (annotation hints)
Scaling a large tool surface
16. Group a large domain into facades
17. Cache the tools/list response
18. Generate tools instead of writing them
Talking to the outside world
19. Query a database from a tool
20. Talk to a TCP socket from a tool
Where to go next

Three layers of authorization
Install
Configuration
The contract
Quick example
Handler interface
@requires rules
Shared types
Tool DSL
Contract validation
Multi-domain routing
Tool grouping (facades)
RBS type syntax
Performance
Caching tools/list
Development
How it works
Stateless transport and schema lifetime
Requirements
License

Declarative tool grouping with summarized facades + deferred schema loading

Requirements

Ruby >= 3.1
Rails >= 6.0
mcp ~> 0.10

Collected from README.md in the repository. Edit it there, not here.

PreviousStateless transport and schema lifetime NextLicense

mcp_authorization v0.7.1 — source · rubygems · MIT licensed