sniffnet/THREAT_MODEL.md

Threat Model

This is Sniffnet's first threat model.
The model is intentionally scoped to a subset of the application — future iterations will expand coverage.

We'll use STRIDE, a widely adopted framework to help identify common types of threats:

• Spoofing: Impersonating someone or something else (e.g., faking identity).
• Tampering: Modifying data or code (e.g., altering a file, changing network packets).
• Repudiation: Denying an action that occurred (e.g., a user denies making a transaction).
• Information Disclosure: Revealing sensitive data to unauthorized individuals (e.g., leaking PII, credentials).
• Denial of Service: Preventing legitimate users from accessing a service or resource (e.g., crashing a server, exhausting resources).
• Elevation of Privilege: Gaining unauthorized access to higher-level permissions (e.g., a regular user becoming an administrator).

For how to report a vulnerability, see SECURITY.md.
For how incidents are handled, see INCIDENT_RESPONSE.md.

For all security-related communications, please write to security@sniffnet.app.

1. Project overview and scope of this document

• Project name: Sniffnet
• Brief description: Cross-platform desktop application to monitor Internet traffic.
• Key features in scope of this model:
  • Packet capture and parsing from a local network adapter or imported PCAP file
  • User-provided files (custom MaxMind MMDB databases, IP blacklists, custom theme palettes)
  • Persisted and reloaded app configurations (settings, favorites, notification preferences, etc.)
  • Remote notifications sent to a user-configured HTTP endpoint

2. Assets to protect

Asset	Why it matters
Host system integrity	Sniffnet processes untrusted inputs (network traffic, user files). A vulnerability could lead to compromise of the user's system.
User's data privacy	Sensitive data (including the captured traffic metadata and custom app configurations) could be exposed to a third party via a vulnerability or misconfiguration.
Application availability	A crash during a monitoring session could cause loss of visibility into the user's network activity at a critical moment.
Project reputation	A vulnerability that leads to compromise of user systems or data would damage trust in the project and its maintainers.

3. Threats (STRIDE)

#	Component / Flow	Spoofing	Tampering	Repudiation	Information disclosure	Denial of Service	Elevation of Privilege
1	Packet capture and parsing	—	T1	—	I1	D1	E1
2	Custom MMDB file	—	T2	—	—	—	—
3	IP blacklist file	—	T3	—	—	D3	—
4	Custom palette file	—	T4	—	—	—	—
5	App configurations file	—	T5	—	I5	—	—
6	Remote notifications POST	S6	—	—	I6	—	—

Threat catalog

• T1 — Malformed packet triggers parser bug. A crafted frame on the wire reaches LaxPacketHeaders::from_ethernet / from_ip / from_ether_type. A logic bug in etherparse (or in Sniffnet's post-parse handling) could produce memory-safety issues, incorrect attribution, or panics.
• I1 — PCAP export discloses captured traffic. When the user enables Savefile, captured packets — headers, hostnames, DNS queries, and payloads — are written to disk unencrypted and can be read by any local process.
• D1 — Packet flood DoS. High packet rates or many small packets force the parser thread to fall behind; GUI freezes or memory grows unboundedly.
• E1 — Privilege escalation via parser RCE. If T1 becomes exploitable, the attacker gets code execution inside a process holding CAP_NET_RAW (Linux) or Npcap driver access (Windows). There is no privilege drop after capture handle creation.
• T2 — Malicious MMDB. A crafted .mmdb targets the maxminddb reader (tree traversal, decoder). open_readfile memory-maps the file, so parser bugs touch privileged memory.
• T3 — Blacklist tampering. Another process running as the user can edit the file loaded by IpBlacklist::from_file to remove entries for known-bad IPs (suppressing alerts) or add benign IPs (driving false positives). There is no signature, checksum, or integrity check on the blacklist contents.
• D3 — Huge blacklist file. IpBlacklist::from_file loads the whole file into memory via tokio::fs::read_to_string — no streaming, no size cap. A multi-GB file would exhaust memory.
• T4 — Palette tampering. Palette::from_file accepts any well-formed hex color without semantic validation. A tampered TOML can set alert colors to be indistinguishable from normal-traffic colors, or render the UI illegibly, causing the user to miss or misattribute security-relevant GUI state.
• T5 — Config tampering. Another process running as the user can edit conf.toml to insert a malicious MMDB path, blacklist path, palette path, or remote webhook URL. On next launch Sniffnet trusts these paths.
• I5 — Config discloses watch list. Favorites, custom file paths, and the webhook URL can be read by any local process.
• S6 — Webhook impersonation. A tampered conf.toml redirects notifications to an attacker's server while the user still believes they're sending to their own bot / SIEM. The JSON body (LoggedNotification::to_json) contains host info, service, favorite metadata, and byte counts — i.e., the user's own network activity.
• I6 — Webhook body discloses monitored assets. The JSON emitted by LoggedNotification::to_json includes favorited host country / ASN / domain and, for blacklist hits, the malicious IP, its reverse-resolved domain, and byte counts. A compromised webhook endpoint — or an on-path observer if the URL is plaintext — learns which assets the user watches and which security events have fired.

4. Mitigations

#	Threat	Mitigation	Status
M1	T1 / E1 / T2	Keep etherparse, pcap, and maxminddb on the latest patched versions.	Active — Dependabot monitors Cargo manifests.
M2	T1 / E1	Add fuzz targets for the packet ingestion entry points (from_ethernet, from_null, from_linux_sll) using cargo-fuzz.	Not implemented.
M3	D1	Bound the pcap channel / parsing queue and drop oldest rather than growing memory unboundedly.	Needs review of pcap_tx backpressure.
M4	D3	Cap the blacklist file size (e.g. 10 MB) before read_to_string.	Not implemented.
M5	I6	Reject non-https:// webhook URLs in the settings UI.	Not implemented.