
Reput.io Integrations on GitHub: Splunk, Sentinel, Elastic, Python and an OpenAPI Spec
We published an MIT-licensed repository with ready-to-use integrations for the Reput.io reputation API: a dependency-free Python client, a Splunk external lookup, Microsoft Sentinel playbook steps, Elastic Security alert enrichment, curl one-liners and an OpenAPI 3.1 spec.
The most common question we get after someone runs their first lookup is a practical one: how do I get this into my SIEM? The API is a single POST call, but every SOC has its own plumbing, and nobody wants to write a client, a retry loop and a batching layer just to try a reputation feed.
So we put the plumbing in public. The repository is github.com/reputio-dev/reputio-integrations, released under the MIT License. Copy what you need, change what you don't like.
What's in the repository
| Folder | What it does |
|---|---|
openapi.yaml | OpenAPI 3.1 description of the Lookup API, also served at reput.io/openapi.yaml |
python/ | A Python client with no third-party dependencies, plus a script that enriches a CSV of indicators |
splunk/ | A Splunk external lookup you can call from any search |
sentinel/ | Logic App playbook steps that add Reput.io context to Microsoft Sentinel incidents |
elastic/ | A script that writes verdicts back onto Elastic Security alerts |
curl/ | One-liners for quick checks from a terminal |
The Python client
python/reputio.py uses only the standard library, so it drops into a SOAR action, a Splunk app or a cron job without a pip install. It handles the parts that usually bite in production:
- Batching. Long lists are split into plan-sized batches (10 indicators per request on Free and Starter, 50 on Pro, 100 on Team and Enterprise).
- Rate limits. On
429it honoursRetry-After;503responses and timeouts are retried with exponential backoff. - Matching. The API returns each indicator trimmed and lowercased, so
lookup_map()keys results by that normalised form. Matching on the raw input string is the classic way to lose a verdict onEvil.Example.COM.
from reputio import Reputio
client = Reputio() # reads REPUTIO_API_KEY
for r in client.lookup(["8.8.8.8", "cloudflare.com", "91.202.233.27"]):
print(r["indicator"], r["status"], r["verdict"])
Need to triage an export? enrich_csv.py adds reputio_status, reputio_verdict, reputio_risk, reputio_provider, reputio_action and reputio_categories columns to any CSV, and deduplicates indicators first so a repeated IP costs one query.
Splunk, Sentinel and Elastic
Splunk. The external lookup turns a reputation check into one line of SPL. Run it after stats so each distinct indicator is queried once:
index=firewall action=allowed
| stats count by dest_ip
| lookup reputio indicator AS dest_ip OUTPUT reputio_status reputio_verdict reputio_provider reputio_action
| where reputio_verdict!="likely_benign"
| sort - count
If the API is unreachable, the script fails soft: rows come back unenriched and the error goes to Splunk's search.log, so a network blip never breaks a search.
Microsoft Sentinel. A no-code Logic App playbook: get the incident's IP, URL or host entities, send them in one HTTP action, and write the verdicts back as an incident comment.
Elastic Security. enrich_alerts.py reads recent alerts, looks up each distinct source.ip and destination.ip once, and stores the result under reputio.* on the alert. When an alert carries several indicators, the worst verdict wins, so a malicious destination is never hidden behind a benign source.
Reading the result
Every response has a status and a verdict. Here's the curl one-liner from the repo against three indicators, run today against production:
8.8.8.8 whitelisted likely_benign info Public DNS Resolver
91.202.233.27 threat malicious high -
cloudflare.com whitelisted investigate medium Cloudflare
The third line is deliberate. Cloudflare is legitimate infrastructure, but anyone can put a site behind it, so a Cloudflare match tells you which network carries the traffic and leaves open who is behind the site. Cloud, CDN and hosting providers stay at investigate with a hint about what to check (Host header, SNI, the actual domain). Only first-party services that don't sell general hosting, like a public DNS resolver, earn likely_benign. And a threat-feed match always wins over a whitelist match, on every plan including Free.
The raw reasons and categories from every matching source come back in the response too, so an analyst or an automation can see exactly why a verdict was reached.
The OpenAPI spec
openapi.yaml describes the request, every response field, the rate-limit headers and the error codes. Import it into Postman or Insomnia, generate a client in your language of choice, or hand it to an AI coding assistant that needs to call the API correctly. It's the same file in the repo and at reput.io/openapi.yaml.
Getting started
- Create a free account at reput.io (500 lookups a day, no card) and copy your API key.
- Clone reputio-dev/reputio-integrations and pick the folder for your stack.
- Check the Lookup API reference for every field in the response.
Using a tool we don't cover yet, like QRadar, Chronicle or a SOAR platform? Open an issue on the repository and tell us what your pipeline looks like. Pull requests are welcome too.
Ready to Try Reput.io?
Start reducing false positives today with our free plan.