Dashboard¶

The Compliance Dashboard provides real-time visibility into the operational status of your AI deployment. It shows call volumes, detection outcomes, and regulatory scorecard metrics.

GET /v1/glad/dashboard¶

Returns the full dashboard payload used by the web interface.

curl http://localhost:8199/v1/glad/dashboard

Query Parameters¶

Parameter	Type	Description
`deployer_id`	`string`	Optional. Filter results to a specific deployer. Defaults to `"default"`.

Response¶

The dashboard response includes:

Call Metrics (runtime):

Field	Description
`total`	Total number of inference calls in the database
`passed`	Calls where all detection axes were below threshold
`prompt_blocked`	Calls where the input prompt was blocked
`answer_unsafe`	Calls where the answer was flagged as unsafe
`answer_safe`	Calls where the answer passed safety checks
`hallucinated`	Calls where the answer was flagged as hallucinated
`grounded`	Calls where the answer was grounded (context faithfulness passed)
`flagged_any`	Calls flagged on any axis

Compliance Status:

Field	Description
`subsystems`	Per-component compliance status (7 subsystems)
`frameworks_compliant`	Number of regulatory frameworks meeting compliance criteria
`kill_switch_active`	Whether the kill switch is currently engaged
`pending_reviews`	Number of calls awaiting human review
`chain_integrity`	Whether the audit chain passes integrity verification

GET /v1/glad/scorecard¶

Returns a regulatory compliance scorecard — a structured summary of which requirements are met, which are pending, and which require attention.

curl http://localhost:8199/v1/glad/scorecard

Query Parameters¶

Parameter	Type	Description
`deployer_id`	`string`	Optional deployer filter. Defaults to `"default"`.

Response¶

The scorecard contains one entry per regulatory framework in your configuration. Each entry shows:

Field	Description
`framework`	Framework code (e.g., `"EU_AI_ACT"`)
`score`	Numeric compliance score (0–100)
`status`	`"compliant"`, `"partial"`, or `"non_compliant"`
`requirements`	List of specific requirements with their individual status
`gaps`	Requirements that are not yet met, with recommendations

GET /v1/glad/health¶

A lightweight health check for the compliance backend. Does not require database access.

curl http://localhost:8199/v1/glad/health

{"status": "ok", "version": "1.0.0"}

GET /health (root)¶

The root health check also returns the active configuration snapshot:

curl http://localhost:8199/health

{
  "status": "ok",
  "version": "1.0.0",
  "config": {
    "model_loaded": true,
    "device": "cuda:0",
    "kill_switch_active": false,
    "applicable_laws": ["EU_AI_ACT", "GDPR", "ISO_42001"]
  }
}