Skip to content

Security: riceharvest/notnet

Security

SECURITY.md

Security Policy — notnet

notnet is a research-purpose botnet framework. It is published so defenders can study, detect, and block this class of malware — and so the offensive capabilities are matched, in the same repository, by shipped detections, intel, and guardrails.

Intended use

  • Security research on systems you own or are explicitly authorized to test.
  • Building/validating the detection artifacts in detections/ and intel/.
  • Cyber-range / detection-engineering exercises with the tests/sim harness.

notnet is not intended for unsanctioned deployment on third-party systems. Doing so is unauthorized access and is illegal in most jurisdictions.

Responsible disclosure

If you find a defect in the defensive tooling — a detection that fails to fire, an intel IOC that has drifted, or a guardrail that does not hold — please report it privately via GitHub Security Advisories (https://github.com/riceharvest/notnet/security/advisories/new). This is the only monitored intake channel; there is no security email address for this repository. If the defect is not sensitive (e.g. a detection gap with no operational impact), a regular issue is also fine.

Response and disclosure timeline (best effort)

This is a research/simulation repository maintained on a best-effort basis; there are no hard SLAs, but the intent is:

  • Acknowledgment: within ~72 hours of filing an advisory.
  • Triage: within ~1 week — confirm/deny, assess impact on detections, intel, or guardrails.
  • Fix: as soon as practical; defensive defects take priority.
  • Disclosure: coordinated with the reporter, defaulting to public disclosure once a fix lands or 90 days after the report, whichever comes first.

If you receive no acknowledgment within two weeks, assume the report was missed and follow up in a public issue (without sensitive details).

Supported versions

Only main is supported: this repo ships no releases and carries a single moving version. Fixes land on main via normal PRs; there are no backports or long-term-support branches. If you run an older checkout, update to the latest main before reporting.

Reporting abuse of leaked builds

A stock notnet binary cannot be disarmed by its operator (#130): the author killswitch domain is baked in at build time. If a build with your author domain leaks:

  1. Point KILLSWITCH_DOMAIN_DEFAULT (the domain you built with) at 127.0.0.1 or 0.0.0.0. Every stock-built instance resolves it and self-destructs.
  2. The killswitch domain is published as a sinkhole-able IOC in intel/ (#152) — defenders can also sinkhole it to neuter leaked stock fleets.

Signed-disarm (a cryptographically signed stay-alive beacon; see #140) makes a leak recoverable: a fork that strips the check also strips signature verification, so the binary will not accept operator commands.

Build attestation

Every release binary should carry a BUILD-ATTESTATION.json (see attest.py) recording the build host, git commit, signer, and the binary's SHA-256. A leaked artifact is traceable to who produced it. CI generates this for CI-produced artifacts.

Scope note (GitHub AUP)

This repository ships an operational bot and the defenses against it. The defenses — detections/, intel/, and this policy — are the argument that the repo is research tooling, not malware for deployment. Keep them in lockstep: do not ship an offensive change without its detection, and do not weaken a guardrail without opening an issue first.

There aren't any published security advisories