PromptArmor

PromptArmor (YC W24) is an LLM application security platform that detects and blocks enterprise-grade threats - indirect prompt injection, data exfiltration, phishing, and system manipulation - in production AI applications. A real-time detection API analyzes LLM inputs and outputs against a continuously updated set of threat detectors before a completion is acted on, returning a fast verdict (for example containsInjection) so applications can block or allow content.

3 APIs 0 Features

AILLMSecurityPrompt InjectionThreat Detection

APIs

PromptArmor Content Check API

Submits content (such as untrusted text an LLM is about to summarize or act on) to PromptArmor's detection engine and returns a verdict - including a containsInjection flag - in...

PromptArmor Analyze API

Analyzes LLM input (for example an email being summarized) and LLM output (for example the generated summary) through PromptArmor's default and modifiable detectors - data exfil...

PromptArmor Session Monitoring API

When a session_id is supplied on analyze requests, PromptArmor runs session-based anomaly detection, learning normal application behavior and flagging deviations across a sessio...

Collections

PromptArmor API

OPEN

Pricing Plans

Prompt Armor Plans Pricing

2 plans

PLANS

Rate Limits

Prompt Armor Rate Limits

2 limits

RATE LIMITS

FinOps

Prompt Armor Finops

FINOPS

Resources

Sources

opencollection: 1.0.0
info:
  name: PromptArmor API
  version: '1.0'
request:
  auth:
    type: apikey
    key: Api-Key
    value: '{{apiKey}}'
    in: header
items:
- info:
    name: Content Check
    type: folder
  items:
  - info:
      name: Check content for adversarial / injected instructions
      type: http
    http:
      method: POST
      url: https://api.promptarmor.com/v1/check_content
      body:
        type: json
        data: "{\n  \"content\": \"<string>\",\n  \"session_id\": \"<string>\",\n  \"detectors\": [\"<string>\"]\n}"
    docs: Submits content to the PromptArmor detection engine and returns a verdict including a containsInjection boolean.
- info:
    name: Analyze
    type: folder
  items:
  - info:
      name: Analyze LLM input
      type: http
    http:
      method: POST
      url: https://api.promptarmor.com/v1/analyze/input
      body:
        type: json
        data: "{\n  \"text\": \"<string>\",\n  \"session_id\": \"<string>\",\n  \"detectors\": [\"<string>\"]\n}"
    docs: Analyzes LLM input and runs the detectors, returning a verdict. session_id enables session-based anomaly detection.
  - info:
      name: Analyze LLM output
      type: http
    http:
      method: POST
      url: https://api.promptarmor.com/v1/analyze/output
      body:
        type: json
        data: "{\n  \"text\": \"<string>\",\n  \"session_id\": \"<string>\",\n  \"detectors\": [\"<string>\"]\n}"
    docs: Analyzes LLM output and runs the detectors, returning a verdict. session_id enables session-based anomaly detection.