API Reference

Get from zero to your first API call in under 3 minutes.

curl -X POST https://zentri.us/api/call/perplexity-search \
  -H "Authorization: Bearer sk-am-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "latest AI news 2025"}'

All requests to the proxy must include your API key in the Authorization header using the Bearer scheme.

Authorization: Bearer sk-am-YOUR_KEY

You can generate and revoke keys anytime from Dashboard → Account. Keys are hashed — we never store the raw value, so save it when you create it.

Every API on the marketplace is called through a single proxy endpoint. Your agent never talks to the upstream provider directly — Zentri handles auth, billing, and rate limiting.

The apiId is the unique slug for each API shown in the marketplace (e.g. perplexity-search). The request body is forwarded as-is to the upstream provider.

Example — Perplexity Web Search

Search the live web and get cited, up-to-date answers. Costs 5 credits per call.

curl -X POST https://zentri.us/api/call/perplexity-search \
  -H "Authorization: Bearer sk-am-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "latest AI models released in 2025",
    "search_type": "pro"
  }'

Every successful response is the original provider JSON with one extra field injected at the top level: _marketplace.

{
  // ... everything the upstream provider returned ...
  "results": [ { "title": "...", "url": "...", "snippet": "..." } ],

  // injected by Zentri
  "_marketplace": {
    "credits_used": 5,
    "credits_remaining": 995,
    "latency_ms": 312,
    "api_id": "perplexity-search"
  }
}

The response also sets three HTTP headers you can read without parsing JSON:

Every API on the marketplace has a base credit cost per call. Some APIs also support variable pricing — credits depend on a parameter in your request body (e.g. video duration, token count).

1 credit = $0.01. Credits are deducted after a successful call. If the upstream provider returns an error, credits are not deducted.

Variable pricing example

An API like Sora 2 (video generation) charges different credits based on the duration parameter:

// 5-second clip → 60 credits
{ "prompt": "A drone flying over the ocean", "duration": 5 }

// 10-second clip → 95 credits
{ "prompt": "A drone flying over the ocean", "duration": 10 }

The exact pricing tiers for each API are shown on its marketplace listing.

Buy credits from Dashboard → Billing. Credits never expire.

Each API enforces two independent rate limits measured in requests per minute (RPM):

When you hit a rate limit, the response is a 429 with a Retry-After header telling you how many seconds to wait:

// HTTP 429 Too Many Requests
// Retry-After: 12

{
  "error": "Rate limit exceeded (per-user RPM)",
  "retry_after_seconds": 12
}

All errors follow the same shape: an HTTP status code and a JSON body with an error field.

Insufficient credits (402)

The 402 response includes extra fields to help your agent recover:

{
  "error": "Insufficient credits",
  "credits_required": 5,
  "credits_available": 2,
  "top_up_url": "https://zentri.us/pricing"
}

Zentri APIs are designed to be used directly by AI agents via .skill.md files. Drop one into your Claude Code project and the agent discovers and calls the tool automatically — no extra integration needed.

Here is the .skill.md for the Perplexity Search API:

---
skill: perplexity-search
description: Search the live web and return cited, up-to-date results. Use this whenever the user asks about recent events, current prices, news, or anything that requires real-time information.
---

# perplexity-search

## Endpoint
POST https://zentri.us/api/call/perplexity-search

## Authentication
Authorization: Bearer <AGENTMARKET_API_KEY>

## Parameters
| Name        | Type   | Required | Description                              |
|-------------|--------|----------|------------------------------------------|
| query       | string | yes      | The search query                         |
| search_type | string | no       | "default" or "pro" (default: "default")  |

## Cost
5 credits per call

## Example
```json
{
  "query": "latest GPT-5 benchmarks",
  "search_type": "pro"
}
```

## Response
Returns the provider response with a _marketplace block appended:
```json
{
  "results": [...],
  "_marketplace": {
    "credits_used": 5,
    "credits_remaining": 995,
    "latency_ms": 312,
    "api_id": "perplexity-search"
  }
}
```

Using with Claude Code

# 1. Set your key
export AGENTMARKET_API_KEY=sk-am-YOUR_KEY

# 2. Drop the skill file in your project root
cp perplexity-search.skill.md ./

# 3. Run Claude Code — it picks up the skill automatically
claude

Claude Code reads .skill.md files from the project root. Once loaded, the agent can call the API on its own when it determines a web search is needed.