Capabilities

Concept: Applications | SDK: Python Configuration

Returns the features available to your API key’s plan: evaluation modes, compliance frameworks, agent and DPDP capabilities, and request limits. Use it to discover the surface area you can build against without hard-coding plan assumptions.

Request

curl https://api.responsibleailabs.ai/railscore/v1/capabilities \
  -H "Authorization: Bearer YOUR_RAIL_API_KEY"

from rail_score_sdk import RailScoreClient

client = RailScoreClient(api_key="YOUR_RAIL_API_KEY")
caps = client.get_capabilities()  # added in SDK 2.6.0; no credits

print(caps.plan, caps.limits.get("requests_per_day"))

Response

{
  "plan": "pro",
  "evaluation": { "modes": ["basic", "deep"] },
  "compliance": {
    "frameworks": ["GDPR", "CCPA", "HIPAA", "EU AI Act", "India DPDP", "India AI Governance"],
    "continuous_monitoring": false
  },
  "agent": { "tool_call": true, "tool_result": true, "prompt_injection": true },
  "dpdp": { "scan": true, "evaluate": true, "evidence": true, "base_path": "/railscore/v1/compliance/dpdp" },
  "limits": { "requests_per_day": 1000, "max_concurrent": 20, "max_active_keys": 5 }
}

evaluation.modes

string[]

Evaluation depths listed here: basic, deep. Note this list doesn’t currently include "auto", even though /eval accepts mode: "auto" — don’t gate auto on this field.

compliance

object

frameworks lists the compliance frameworks you can evaluate against. continuous_monitoring indicates whether ongoing monitoring pipelines are available on your plan.

agent

object

Agent governance features: pre-call tool_call checks, post-call tool_result scanning, and standalone prompt_injection detection.

dpdp

object

India DPDP features available on your plan: scan, evaluate (generally available), and evidence (Pro tier and above — audit-grade evidence packets). base_path is the DPDP endpoint prefix.

limits

object

Request limits for the key: requests_per_day (null means unlimited), max_concurrent, and max_active_keys.

Reference

Scoring

Compliance

Agent

Introspection & status

Request

Response

​Request

​Response

Request

Response