in short

  • static, read-only analysis of public builds. no exploitation, no user data, no disruption.
  • private report to the owner or the national cert first. nothing public while a flaw is live.
  • secrets are never published, anywhere, ever.
  • we re-verify fixes and let the owner or cert lead on timelines.

1. our commitments

  • we perform static, read-only analysis of publicly available applications, obtained the way any user would obtain them.
  • we verify every finding by hand and describe only what we observed, deferring impact and severity to the owner.
  • we report privately to the affected owner, or to the relevant national cert where the owner is a government or public body, before any public mention.
  • we never publish secrets, credentials or private keys. full evidence is shared with the owner over a secure channel, and we will sign an nda where appropriate.
  • we give reasonable time to remediate, extend it when an owner is engaged, and coordinate any public communication with the owner or cert.
  • we never ask for payment to withhold, delay or soften a report. if an owner runs a bounty program we follow its rules; if it chooses to reward us, we are grateful; neither changes what we report.

2. what we look at

our pipeline examines android application packages (apks) that are publicly distributed. it looks for secrets that should never ship in a client: hardcoded credentials, live cloud and payment keys, oauth client secrets, bundled private keys and certificates, and tokens that grant access to backend services. it also records what an app can do and where it sends data, for the behaviour work that feeds monkguard.

we do not test live services, web applications or infrastructure, and we do not attempt to log in, authenticate or use any credential we find. confirming that a secret is live is done in the least intrusive way that proves exposure, and no further.

3. what we never do

  • exploit a finding, or go beyond what is needed to confirm exposure.
  • access, download or retain real user data.
  • disrupt, degrade or load-test a live service.
  • name an affected app, organisation or developer while a flaw is live.
  • publish a credential, key or anything that could be used to exploit an issue.
  • overstate impact, or present a finding as more certain than it is.
  • share a report with anyone other than the owner or the relevant cert.

4. the timeline

dates are targets, not threats. an engaged owner with a fix in progress will always get more time.

  1. 01
    day 0

    we find and confirm it

    a candidate from the scanner is verified by hand: live, real, and belonging to this owner. only then does it exist as a finding.

  2. 02
    day 0 to 3

    we report privately

    to the owner’s security contact, or to the national cert for public bodies. evidence is redacted in the first message; full detail over a secure channel.

  3. 03
    day 7

    we follow up

    if we have heard nothing, we try a second route: a published security.txt, a bounty program, a cert, a dpo.

  4. 04
    day 30

    we check in

    a courteous reminder with the same reference, and an offer to help with rotation or to clarify.

  5. 05
    day 90

    a fix is expected

    the industry-standard window. we extend it readily when an owner is engaged and the fix is in progress.

  6. 06
    after the fix

    we re-verify and close

    we download the new build, confirm the secret is gone, and close the case with the owner or cert. an anonymised case study may follow, with agreement.

if, after repeated attempts over at least 90 days, an owner cannot be reached or declines to engage, we may ask the national cert of the owner’s country to assist. we still do not publish details of a live flaw. in the rare case where silence puts a large number of people at ongoing risk, we will publish only the minimum needed for users to protect themselves, after notice, and never a secret.

5. what a report contains

sectionwhat it says
summaryone paragraph: what we found, in which app and version, and why it matters.
evidencewhere the secret lives in the build, a redacted copy, and how you can confirm it yourself.
exposurewhat the secret grants, as far as we could confirm without using it.
remediationrotate, remove from the client, move to the server, and what else to check.
our commitmentsa copy of this policy and a secure channel for the full evidence.
referencea mashmonk reference, and the cert’s reference where one exists, on every message.

6. what we publish

our public research pages show only anonymised, aggregated information: sector, country, install band, vulnerability class, severity and status. we never name an affected application, organisation or developer, and we never publish anything that could identify or exploit an issue that is not yet fixed. a fixed issue may become an anonymised case study, with the owner’s agreement and in the form they agree to.

7. if you are an organisation

if you have received a report from us, thank you for engaging. reply on the same thread for clarification, for the full evidence over a secure channel, or to confirm a fix; we will re-verify on the new build and close the case with you. if you would like us to review one of your apps proactively, before or after release, get in touch. an nda is fine.

we are happy to be credited, and equally happy not to be. tell us which you prefer.

8. if you are a national cert

we work with national certs as a research partner. we route public-sector findings through your intake, follow your handling and encryption requirements, keep your reference number on all correspondence, and let you lead on timelines and on contact with the affected body. where you prefer we contact the body directly and copy you, we will do that instead.

9. if you are a researcher

if you have found something in a mobile app and would like help reporting it responsibly, write to us. we will verify it, report it under this policy, and credit you if you wish. if you would like to work with us, we would like to hear from you.

10. safe harbour

we intend our own research to be lawful, good-faith and non-disruptive, and we conduct it in line with widely accepted coordinated-disclosure norms (including iso/iec 29147 and 30111). we ask organisations to treat reports sent in this spirit as authorised security research and not to pursue action against good-faith reporting. we extend the same courtesy to anyone who reports to us in good faith.

11. reaching us

security contact: hello@mashmonk.com · signal: @devzoy. a pgp key for encrypted evidence is available on request and referenced in /.well-known/security.txt. please do not include live secrets in a first message.