MISP Threat Intelligence
A POC evaluating MISP as an IOC source for WAF/CDN/SOC context — explicitly scoped as a proof of concept, not a production integration.
Why
The target environment runs a WAF, a CDN, and a SOC. Each of these surfaces benefits from a shared, current view of known-bad indicators — malicious IPs, domains, URLs, and file hashes — so that detection and enrichment draw on the same external knowledge instead of isolated, ad-hoc lists.
MISP (Malware Information Sharing Platform) is the open-source standard for collecting, normalizing, and sharing threat intelligence. The gap under evaluation is whether a self-hosted MISP instance, fed by public OSINT sources, produces indicators good enough and fast enough to serve as an upstream IOC source for that ecosystem. This research answers that question with evidence before any production commitment.
The work also maps directly to ISO 27001 controls A.5.7 (Threat Intelligence) and A.5.24 (Incident management readiness): standing up MISP and observing its feeds generates the documented evidence those controls require.
Scope
- Deployment: MISP core plus modules via
docker compose, running locally on Docker Desktop (~4 vCPU / 8 GB RAM), not exposed publicly. - Feeds: four free, no-commercial-license public sources — CIRCL OSINT (curated hourly), abuse.ch URLhaus, ThreatFox, and Feodo Tracker.
- Observation: a 7-day window collecting daily counts of events, attributes, objects, and database growth.
- Consumption: REST API access through PyMISP to pull IOC lists and measure latency and output format.
- Out of scope: production/VM deployment, live integration into the WAF, SAML/OIDC SSO, and clustering or HA — each deferred to a later phase.
Problems & Solutions
Signal versus noise in feed selection
MISP ships ~70 default feeds; enabling all of them inflates the database and drowns useful indicators. The research enables exactly four high-trust sources and keeps the rest off, trading breadth for a clean, auditable dataset.
Reproducible feed enablement
Enabling feeds by hand through the web UI is not repeatable. A scripts/enable-feeds.sh helper drives the REST API to load feed metadata, enable the chosen feeds by UUID, and trigger the first fetch — making the setup scriptable and repeatable.
Fitness as a WAF/CDN feed source
The open question is whether MISP output can drive a rule engine. The REST API is exercised via PyMISP and curl to pull IOC lists, measure API latency, and confirm the JSON/CSV/STIX output and tag/type/date filtering match downstream input expectations.
Architecture
Redis-backed workers fetch and cache the four feeds on a schedule (fetch every 6h, cache every 12h). Indicators land in the MISP core database, where PyMISP and the REST API expose them as filterable IOC lists for the WAF, CDN, and SOC to consume.