safety
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| SAFETY_CONFIG | No | Path to a JSON configuration file. Used by the server (via --config) to load settings; omitted values fall back to defaults. | |
| SAFETY_PROVIDER_KEY | No | API key for the model-backed provider used by RealProvider. | |
| SAFETY_PROVIDER_URL | No | URL of the model-backed provider used by RealProvider. If unset, RealProvider raises ProviderUnavailable. | |
| SAFETY_MAX_INPUT_CHARS | No | Overrides the maximum number of characters allowed in input payloads without editing the config file. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| screenC | Classify an input or output payload for a tenant and return an allow/block/review decision. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 1 tool
Only one tool exists, so there is no possibility of confusing it with another tool.
The single tool name 'screen' is a clear, action-oriented verb and introduces no naming inconsistencies.
A single screening tool feels thin for a 'safety' server, though it does cover a core classification action.
The tool covers the main screening/classification task, but lacks related operations such as policy management or decision history, leaving notable gaps.