# Clari Rate Limits

> Handle Clari API rate limits with backoff and export job scheduling. Use when hitting 429 errors, optimizing export frequency, or scheduling bulk forecast exports. Trigger with phrases like "clari rate limit", "clari 429", "clari throttle", "clari api limits".

- Skill: `gabrielmoreira/clari-rate-limits` (Agent Skill)
- Install (CLI): `npx skillmds@latest add gabrielmoreira/clari-rate-limits`
- Raw SKILL.md: https://api.skillmd.com/api/skills/gabrielmoreira/clari-rate-limits/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Integrations & APIs
- License: MIT
- Author: gabrielmoreira (https://skillmd.com/u/gabrielmoreira)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/gabrielmoreira/clari-rate-limits

---

# Clari Rate Limits

## Overview

The Clari API enforces rate limits per API key. Export jobs are asynchronous and queued server-side, so the primary concern is polling frequency and concurrent export requests.

## Prerequisites

- A scoped API token and approved forecast export scope
- Persistent job/attempt tracking for retry and idempotency decisions
- A scheduler that can defer work without spawning duplicate workers
- Monitoring for queue depth, response class, and terminal failures

## Rate Limit Behavior

| Aspect | Value |
|--------|-------|
| Scope | Per API key |
| Response on limit | HTTP 429 |
| Export job queue | Server-managed, async |
| Recommended polling | 5-10 second intervals |

## Instructions

### Exponential Backoff for Export Polling

```python
import time
import requests

def poll_with_backoff(
    job_id: str,
    api_key: str,
    max_attempts: int = 60,
    base_delay: float = 5.0,
    max_delay: float = 60.0,
) -> dict:
    for attempt in range(max_attempts):
        resp = requests.get(
            f"https://api.clari.com/v4/export/jobs/{job_id}",
            headers={"apikey": api_key},
        )

        if resp.status_code == 429:
            retry_after = int(resp.headers.get("Retry-After", base_delay))
            time.sleep(retry_after)
            continue

        resp.raise_for_status()
        status = resp.json()

        if status["status"] in ("COMPLETED", "FAILED"):
            return status

        delay = min(base_delay * (1.5 ** attempt), max_delay)
        time.sleep(delay)

    raise TimeoutError(f"Job {job_id} did not complete in {max_attempts} attempts")
```

### Sequential Export Scheduler

```python
def export_all_periods(
    client,
    forecast_name: str,
    periods: list[str],
    delay_between: float = 10.0,
) -> list[dict]:
    results = []
    for period in periods:
        print(f"Exporting {period}...")
        job = client.export_forecast(forecast_name, period)
        result = poll_with_backoff(job["jobId"], client.config.api_key)
        results.append(result)
        time.sleep(delay_between)  # Avoid hitting rate limits
    return results
```

## Error Handling

| Scenario | Detection | Response |
|----------|-----------|----------|
| 429 with Retry-After | Check header | Wait exact duration |
| 429 without header | Status code only | Backoff from 5s |
| Job queue full | Multiple pending jobs | Wait for completion before new exports |

## Output

Return a per-job state record with export period, provider job ID, attempts,
applied delay, response class, and terminal decision. Do not report tokens or
raw forecast records; callers must receive a bounded retryable failure rather
than silently starting a parallel export.

## Examples

Submit one export for a staging period and persist its job ID before polling.
On a 429 with `Retry-After`, reschedule the same job after that interval; if
the attempt budget expires, mark the run unavailable and alert the scheduler
instead of issuing another export request.

## Resources

- [Clari API Reference](https://developer.clari.com/documentation/external_spec)

## Next Steps

For security configuration, see `clari-security-basics`.