Clari performance tuning

Skill jeremylongshore/claude-code-plugins-plus-skills/plugins/saas-packs/clari-pack/skills/clari-performance-tuning

425 plugins, 2,810 skills, 200 agents for Claude Code. Open-source marketplace at tonsofskills.com with the ccpi CLI package manager.

Install

npx -y skills add jeremylongshore/claude-code-plugins-plus-skills --skill clari-performance-tuning

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

'Optimize Clari API performance with caching, batch exports, and data pipeline efficiency.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.9 KB, as published. Nobody here has run it

Clari Performance Tuning

Overview

Optimize Clari export pipelines: reduce export times, cache forecast data, and parallelize multi-period exports.

Instructions

Parallel Multi-Period Export

from concurrent.futures import ThreadPoolExecutor, as_completed

def parallel_export(
    client,
    forecast_name: str,
    periods: list[str],
    max_workers: int = 3,
) -> dict[str, list[dict]]:
    results = {}

    def export_period(period: str) -> tuple[str, list[dict]]:
        data = client.export_and_download(forecast_name, period)
        return period, data.get("entries", [])

    with ThreadPoolExecutor(max_workers=max_workers) as executor:
        futures = {
            executor.submit(export_period, p): p for p in periods
        }
        for future in as_completed(futures):
            period, entries = future.result()
            results[period] = entries
            print(f"  {period}: {len(entries)} entries")

    return results

Cache Export Results

import json
import hashlib
from pathlib import Path
from datetime import datetime, timedelta

class ExportCache:
    def __init__(self, cache_dir: str = ".cache/clari", ttl_hours: int = 4):
        self.cache_dir = Path(cache_dir)
        self.cache_dir.mkdir(parents=True, exist_ok=True)
        self.ttl = timedelta(hours=ttl_hours)

    def _key(self, forecast: str, period: str) -> str:
        return hashlib.md5(f"{forecast}:{period}".encode()).hexdigest()

    def get(self, forecast: str, period: str) -> list[dict] | None:
        path = self.cache_dir / f"{self._key(forecast, period)}.json"
        if not path.exists():
            return None
        meta = json.loads(path.read_text())
        cached_at = datetime.fromisoformat(meta["cached_at"])
        if datetime.utcnow() - cached_at > self.ttl:
            return None
        return meta["entries"]

    def set(self, forecast: str, period: str, entries: list[dict]):
        path = self.cache_dir / f"{self._key(forecast, period)}.json"
        path.write_text(json.dumps({
            "cached_at": datetime.utcnow().isoformat(),
            "entries": entries,
        }))

Incremental Load to Warehouse

-- Use MERGE for incremental updates instead of full reload
MERGE INTO clari_forecasts AS target
USING staging_clari AS source
ON target.owner_email = source.owner_email
   AND target.time_period = source.time_period
   AND target.forecast_name = source.forecast_name
WHEN MATCHED THEN UPDATE SET
    forecast_amount = source.forecast_amount,
    quota_amount = source.quota_amount,
    crm_total = source.crm_total,
    crm_closed = source.crm_closed,
    exported_at = source.exported_at
WHEN NOT MATCHED THEN INSERT VALUES (
    source.owner_name, source.owner_email, source.forecast_amount,
    source.quota_amount, source.crm_total, source.crm_closed,
    source.adjustment_amount, source.time_period,
    source.exported_at, source.forecast_name
);

Performance Benchmarks

Optimization	Before	After
Sequential 4-period export	2 min	40s (parallel)
Cache hit	5-10s API call	<1ms
Full table reload	30s	5s (MERGE)

Resources

Clari API Reference

Next Steps

For cost optimization, see clari-cost-tuning.