Description

Thanks for your interest in the 1Password bug bounty program! External security evaluations are an important step on our journey to make (and keep) 1Password the best and most secure password manager on the market. Security is deeply integrated into the architecture of 1Password. The platform is designed around end-to-end encryption, client-side key generation, two-secret key derivation, Secure Remote Password (SRP) authentication, and multiple layers of cryptographic, server-side, and client-side access control. This isn't an easy program. Scanners are unlikely to help, and standard XSS-type injections won't yield much either. We're looking for creative researchers who are willing to understand the security model, think critically about its assumptions, and identify situations where the actual behavior of the product violates those assumptions. We're happy you're here. Please note: Program information for the $1 million Capture the Flag (CTF) Challenge is specific and outlined in the $1,000,000 Challenge section under In scope.

Bounties
Low
0.1 - 3.9
Medium
4.0 - 6.9
High
7.0 - 8.9
Critical
9.0 - 9.4
Exceptional
9.5 - 10.0
Tier 2
min. $
max. $
50
300
300
600
600
6,000
6,000
15,000
15,000
30,000
Tier 2
$50 - $30,000
Rules of engagement
Required
Mozilla/5.0 Intigriti-Researcher/<username>
Not applicable
X-Intigriti-Username: <username>

By participating in this program, you agree to:

  • Respect the Community Code of Conduct
  • Respect the Intigriti Terms and Conditions
  • Respect the defined scope of the program
  • Only interact with accounts, vaults, and resources you own or for which you have explicit permission from the account holder
  • Submit one vulnerability per report, unless you need to chain vulnerabilities to provide impact
  • Stop testing once sufficient impact has been demonstrated
  • Make a good faith effort to avoid privacy violations, destruction of data, and interruption or degradation of our service
  • Avoid accessing, modifying, or retaining another user's data beyond the minimum required to demonstrate impact
  • Avoid denial-of-service, brute-force activity, or any action that could negatively affect other users
  • Keep automated requests/scanning under 45 requests per minute
  • Contact us before running tests that may cause a spike in errors or disrupt service, so we can discuss other options
  • Include a full description, proof of concept, and complete reproduction steps in the initial report, with enough evidence for 1Password and Intigriti to validate the issue. Only detailed reports with reproducible steps are considered valid and eligible for reward.

Important rate-limit note: Automated requests/scanning must be kept to under 45 requests per minute. This is equivalent to an average of 0.75 requests per second, so researchers should keep testing below 1 request per second. Scanners (and anything that sends an excessive number of requests) will add wait time to your tests due to the rate limiting that is in place.


Temporary Blocks

1Password applications are designed with multiple layers of security measures. Intentionally bypassing established measures generally leads to a temporary block from our services, for approximately 24 hours. We typically enforce suspensions for their defined periods, as policy dictates; particularly suspensions that result from conduct that violates our program rules.

We acknowledge there may be situations in which a researcher feels a block has been implemented unjustly — during the course of regular application use, for example. In such cases, we encourage you (the researcher) to contact us via email.

Although we cannot guarantee the removal of your suspension, communicating your concerns will ensure your case is heard as each scenario is evaluated individually.

We appreciate your understanding and help maintaining the integrity and security of 1Password.


Scope Questions and Flaw Hypotheses

  • Our security team does not make scope or eligibility determinations outside of a submitted report.
  • Good-faith flaw hypothesis findings can be submitted to our bug bounty program without penalty. Note them explicitly as flaw hypotheses. They will be closed as Informative.
  • Attempts to abuse this, including fabricated or unverified research or claims about behavior that wasn't actually observed (AI-generated or otherwise), will be treated as spam.

Duplicates

  • The first valid report we receive will be rewarded in the event of duplication. This includes vulnerabilities already known to 1Password from internal testing or from a previous valid submission through any of our reporting channels.
  • Multiple vulnerabilities caused by a single issue will be awarded one bounty.

Rewards and Severity

Our rewards are based on impact to account, user, vault, or item security with reports compromising multiple users on a wider scale increasing severity (e.g.; a local attack with malware is considered lower severity over a remote attack through our web application, while local attacks that can bypass authentication are higher severity than those that require 1Password to be unlocked to exploit). Please note these are general guidelines, and reward decisions are up to the discretion of 1Password.

Rating severity accurately: Please rate severity based on the impact your report actually demonstrates, not the worst case it could theoretically lead to.

Some disagreement on severity is normal. If we assess a finding differently, we'll adjust the rating during validation. An honest misjudgment doesn't count against you or against a valid finding.

Misrepresentation is different. This includes claiming impact the report doesn't demonstrate, describing results that weren't actually observed, or inflating severity to draw attention to a report. We take this especially seriously for Critical and Exceptional submissions. Reports that appear to deliberately misrepresent severity or impact may be treated as spam. Any valid issue in the report will still be assessed at its correct severity.

Leakage of sensitive personally identifiable information: We will consider a vulnerability in this class to be considered "Critical" in cases where 1Password is solely responsible for the security of the data. Often our users can accidentally leak their own data through various means including, but not limited to, uploading Emergency Kits on publicly shared systems, utilizing tools such as VirusTotal to scan emails, and similar. Additionally it should be noted that users are responsible for the security of the devices they trust when signing into their accounts, and 1Password performs a best effort approach to local threats.

Vulnerabilities involving a self-sign-up flow: Self sign-up is possible for net new accounts, but not possible for users being added to existing accounts. We will consider "Privileges Required" to be set to "None" in cases where a vulnerability is exploitable without logging into the victim's 1Password account or if a local attack requires no privileges on the device to exploit.


Disclosure

Please do not discuss this program or any vulnerabilities (even resolved ones) outside of the program without express consent from the organization. This includes all submitted vulnerabilities (duplicate, not applicable, etc.) and PoCs published on YouTube, Vimeo, or other public platforms.


Validation Times

We will aim to validate all submissions within the timelines below once your submission has been verified by Intigriti.

Submissions validated outside of these timelines may be awarded a $25 bonus, at 1Password's discretion.

Vulnerability Severity Time to validate
Exceptional 2 working days
Critical 2 working days
High 5 working days
Medium 15 working days
Low 15 working days
Assets
1
1Password Accounts & Web Application
1
1Password API
1
1Password Browser Extension
<your-own-1password-account> → Latest Browser Extension
Other
Tier 2
1
1Password Command Line Interface
4
Out of Scope Assets
*.agilebits.com
Wildcard
Out of scope
In scope

Get Started

Start with the attached 1Password Security Design White Paper, and pay particular attention to the section titled Beware of the Leopard (page 68). It explains the decisions and considerations behind the 1Password security design. We've also created a tool to help you investigate 1Password requests and responses with your own session key.

Before testing, please also review the attached In-Scope Documentation. The White Paper describes the architecture and security model. The In-Scope Documentation gives practical guidance on where to focus, how to structure the attacker story, and which product behaviors are known and intended.

You may create a trial account for testing. Please sign up using your Intigriti email alias (<username>@intigriti.me) so we can tie your test activity to your Intigriti identity. Download the latest stable version of 1Password or find the Beta versions detailed in our release notes.

If you're interested in testing our nightly build, you can install the nightly release as follows:

  1. Open and unlock 1Password.
  2. Click your account or collection at the top of the sidebar and choose Settings.
  3. Click Advanced, then set "Release channel" to Nightly.

Updates will be installed automatically when "Install updates automatically" is turned on.

Note: Issues found in nightlies may be evaluated differently than issues found in a stable release.


Priority Areas

Research is particularly encouraged around:

  • End-to-end encryption and key management
  • Account Password, Secret Key, and Two-Secret Key Derivation
  • Authentication and Secure Remote Password
  • Vault and item authorization
  • Account and vault isolation
  • Passkeys, SSO, and device linking
  • Account recovery
  • Item sharing
  • Secrets Automation and Connect Server security
  • Cryptographically and server-enforced security boundaries

A strong report should clearly explain:

  1. What capability the attacker already has
  2. What security boundary exists before exploitation
  3. What new capability the attacker gains
  4. Which intended 1Password security property is violated

Access-Control Model

1Password uses a combination of:

  • Cryptographically enforced controls
  • Server-enforced controls
  • Client-enforced controls

Not every restriction shown by a client is intended to provide a cryptographic security boundary. Client-enforced policy can be circumvented by a malicious client or determined user.

Bypassing a client-enforced restriction alone is not sufficient unless the attacker gains a capability or access they were not already entitled to.


LLM Prompt / AI-Assisted Research

This section is meant to provide guidance for AI-assisted research. In fact, you're welcome to paste the reasoning here into your agent or project instructions. Below are the most consistent patterns we see when AI-assisted reports are considered out of scope.

Analyze the attacker story. Before you submit, ask whether the conditions required for the attack already give the attacker what they're after. Common examples of this include scenarios where, prior to exploiting the described vulnerability, an attacker has already gained local code execution, a vault or vault item, JavaScript on a 1Password-owned origin, a phished session, etc. We see this pattern frequently in the AI-assisted reports we close, and avoiding it often requires human analysis and critical thinking.

Your model will agree with whatever frame you give it. LLMs don't verify claims against our product — they predict plausible-sounding next tokens, and they're easy to steer. If you shape the conversation toward "this is in scope" or "this has impact," the model will dutifully build that case. That doesn't make the case real. Treat anything the model tells you about 1Password's threat model, scope, or impact as a hypothesis to verify against the Security Design White Paper, the In-Scope Documentation, this brief, and the actual product behavior — not as a conclusion. If your model says something is out of scope, that's the moment to think hard about why before pushing back on it.

You're accountable for what you submit. Researchers who do well incorporate bespoke problem-solving, creativity, and reproduction against the live product. AI assistance can absolutely help with that, but it can also generate polished noise and evidence that comes from a model's description of 1Password rather than from 1Password itself. When submitting an AI-assisted report, we recommend including a screen recording of the end-to-end exploit where possible. We're here to incentivize and reward valid vulnerability disclosure, and you're the first line of defense for your reputation points.


The $1,000,000 Challenge

We introduced a $1 million CTF bug bounty challenge in 2022 to further our commitment to providing an industry-leading security platform for individuals, families, and businesses. There is exactly one way to earn our top reward: capture the flag from the target vault. The flag is a real value stored in a 1Password vault you do not hold the keys to. If you can produce it, you have broken 1Password's encryption — and we will pay $1,000,000 USD.

The target:

  • Account/vault: h2ied2pqg7qo4ysngam25ic4fm
  • Item IDs: 6og4d3pkva5y2ohexjdtkxpgqe

You attempt this on your own, with your own tooling, on your own time, against the same 1Password that protects our customers. There is no longer a separate account to be invited to and no separate program to join.

What this challenge is not — please read before submitting:

  • It is not designed to be solvable. We believe it isn't, which is the point. The reward exists because we trust the math behind 1Password's security model.
  • It is not an educational or practice CTF. No levels, no hints, no categories, no intended "solution." It is a standing, open bet against our cryptography.
  • It is not for general research. Web bugs, app bugs, and everything else belong in the main program, under our standard reward tiers. This challenge is only the flag.
  • A valid submission must contain the actual captured flag. Not a theoretical path, not a partial chain, not "this should work." The first thing we check is whether you have the real flag. Reports that claim to have captured the flag but do not contain the flag will be marked Not Applicable.
  • The target is the encryption, not the people. Phishing the vault owner, social engineering 1Password staff, or attacking out-of-scope infrastructure does not count.

How to submit: Reports must be made through this program. If you've successfully accessed the flag from the target vault, put the captured flag at the very top of your report.

1Password_Security_Design.pdf
11/13/2026, 5:12:57 PM
1Password_In-Scope_Documentation.pdf
11/13/2026, 5:12:57 PM
Out of scope

Known Issues (16/11/2026) [DD/MM/YYYY]

1Password is aware of the following behaviours. Reports describing the same behaviour and impact may be closed as duplicates.

  • 1Password CLI - deleted items remain recoverable. Items deleted using op item delete can still be recovered.
  • 1Password browser extension - inline menu height can reveal whether certain vault items exist. A webpage can measure the height of the extension’s inline menu to infer whether the vault contains entries such as identities or email addresses. 1Password accepts this risk for the remaining affected fields. The payment card case has been fixed.
  • Other out-of-scope cases. Additional issues may fall outside the program’s scope. We’ll assess these reports case by case against the published scope and the security impact demonstrated. If your finding relates to a known issue, please explain what additional impact you’ve identified so triage can assess it.

All reports for out of scope findings will be closed as Out of Scope and may affect your reputation. Please review our out of scope findings carefully.

When reporting vulnerabilities, please consider (1) attack scenario / exploitability, and (2) security impact of the bug.

In order to be a triaged issue a submission must demonstrate an impact that can have an effect on our users. Submissions should always answer the question "as an attacker I could", with a suitable demonstration of such. Findings that disclose points of information or security best practices without an impact are not eligible for a reward and should be explored further in order to demonstrate risk.

1Password-Specific Exclusions

We may reject reports that meet these criteria, or downgrade their priority:

  • Findings caused solely by scheduled or planned infrastructure changes
  • Attacks that require root access to the victim's device
  • Findings that require direct memory access or dynamic instrumentation
  • Findings caused solely by operating system virtualization or emulation
  • Findings that depend on memory dumps or tools that read active or cached process memory
  • Findings that do not compromise or meaningfully affect account, user, vault, item, authentication, authorization, or other protected security boundaries
  • Findings that only reproduce behavior already documented as an intended limitation in the 1Password Security Design White Paper, unless materially new impact is demonstrated
  • Placeholder or templated reports (e.g. no detailed replication steps or templated information in the initial report). All valid findings must be submitted with a full description, proof of concept, and complete replication steps in the initial submission.

Commonly Reported Issues

These issues have been reviewed numerous times and reached the conclusion that the product design is what we want it to be, or that they are inapplicable based on the published White Paper. They remain out of scope unless a report demonstrates behavior beyond the documented limitation or a materially new security impact.

The reasoning for each is in the attached Commonly Reported Issues document; please read it before submitting a report in any of these categories.

  • SPF/DKIM and other email forgery protection
  • Disclosure of Session Tokens and UUIDs
  • Locally exposed Secret Keys
  • Browser-specific "Local Storage" Use for Secret Keys
  • Rate limiting
  • Bypassing Rate limiting with IP rotation
  • Ability to see content after Account Password change
  • No prompt for Secret Key after deauthorizing a mobile device
  • Known limitations with allowing offline syncing
  • Client-based Access Controls
  • URLs, Emails, and Tokens (or other data) exposed in Wayback Machine or similar third-party tools
  • Weak Password Policy
  • Disclosed Banlist ("/banlist/combined_words.txt")
  • Hyperlink Injection in Emails
  • Billing-related issues or ways to extend trials, use frozen accounts, etc
  • data-fcm-api-key exposure
  • Multifactor Authentication
  • Uploaded Files - No scanning for "malicious" files
  • Keeping and securing the emergency kit
  • Browser Autofill Security
  • Unclaimed NPM Packages
  • Revealing who is registered
  • Item Content Injection
  • CSV Injections

Security must be balanced with usability. We'll always consider feedback about design decisions but ask that you understand 1Password has been extensively reviewed by our internal team and external audits.

AI-Assisted and Theoretical Findings

  • AI-generated vulnerability theories that have not been independently reproduced against the live 1Password product
  • Reports based solely on an LLM's interpretation of the product, threat model, or Security Design White Paper
  • Findings where the attacker's prerequisites already provide the capability the report claims to gain
  • Theoretical security issues without a realistic attack scenario, reproducible exploit path, or demonstrated security impact
  • Reports that describe expected product behavior as a vulnerability without demonstrating a violation of an intended 1Password security property

Good-faith flaw hypotheses are handled separately. See "Scope Questions and Flaw Hypotheses" in the Rules of Engagement.

Application

  • WordPress username disclosure
  • Self-XSS that cannot be used to exploit another user
  • Verbose error messages, files, or directory listings without sensitive information disclosure
  • CORS misconfiguration on non-sensitive endpoints
  • Missing cookie flags
  • Missing security headers
  • CSRF with no or low security impact
  • Presence of autocomplete attributes on web forms
  • Reverse tabnabbing
  • Missing or bypassable rate limits without demonstrated security impact
  • Password policy or other security best-practice observations, including complexity, expiration, and reuse
  • Clickjacking without proven impact or requiring unrealistic user interaction
  • CSV injection without demonstrated security impact
  • Session-management observations without demonstrated impact, including sessions remaining valid after logout or security-setting changes
  • Tokens, identifiers, or UUIDs disclosed where possession does not provide additional authorization or access
  • Email spoofing, SPF, DKIM, or DMARC configuration findings
  • Content or hyperlink injection without meaningful security impact
  • Username or email enumeration without a broader exploit or ability to enumerate users at scale
  • Email bombing
  • HTTP request smuggling without demonstrated security impact
  • Homograph attacks
  • XML-RPC enabled
  • Banner grabbing or version disclosure
  • Metadata not being stripped from uploaded files
  • Same-site scripting
  • Subdomain takeover reports without demonstrating control of the affected resource
  • Arbitrary file upload without demonstrating that the uploaded file exists, is reachable, and creates security impact
  • Blind SSRF without demonstrated business or security impact
  • Disclosed or misconfigured third-party API keys that do not provide access to sensitive backend resources
  • Host header injection without demonstrated business or security impact
  • Known vulnerable libraries or dependencies without proof that the vulnerability is exploitable in the affected 1Password implementation

General

  • Spam, social engineering, phishing, vishing, smishing, or physical intrusion
  • DoS, DDoS, resource exhaustion, email bombing, or brute-force attacks
  • Automated testing exceeding 45 requests per minute
  • Testing likely to cause a significant spike in errors, service degradation, or disruption without prior approval
  • Vulnerabilities that only affect software versions that no longer receive security updates
  • Attacks requiring physical access to a victim's device, man-in-the-middle positioning, local malware, or an already-compromised user account, unless the report demonstrates a security boundary that should still hold under that threat model
  • Testing 1Password accounts that you do not own and for which you do not have explicit permission
  • Accessing another user's vault, account, item, or data beyond the minimum required to demonstrate impact
  • Destructive testing, modification or deletion of data, or activity that risks service interruption
  • Recently disclosed zero-day vulnerabilities affecting in-scope assets within 14 days of a public patch or mitigation release are generally not eligible for a bounty
  • Reports that only state that software is outdated or vulnerable without a working proof of concept and demonstrated impact
  • Public disclosure of program information or vulnerabilities without prior written approval
  • Findings that require disabling normal security controls or materially weakening the researcher's own environment before exploitation can occur, unless the same condition can realistically affect another user
1Password - Commonly Reported Issues.pdf
11/13/2026, 6:13:55 PM
Severity assessment

This program follows Intigriti's triage standards based on the proof of concept.

FAQ

Can I create a 1Password account specifically for testing?

Yes. Researchers may self-register and create a trial 1Password account for testing. Please sign up using your @intigriti.me email address (see below).

Only test accounts you own or accounts for which you have explicit permission.


Do I need to use an @intigriti.me email address?

Yes. Please use your Intigriti email alias (<username>@intigriti.me) when signing up for any 1Password account you use for testing. This helps us recognize legitimate research activity and connect test accounts to your Intigriti profile.


Can I test Family or Business accounts?

Yes, provided the account is owned or controlled by you, or you have explicit permission from the account owner.


Can I test another user's account?

No, unless you have explicit permission from the account owner.

Do not access another user's account, vault, item, or data beyond the minimum necessary to demonstrate a valid vulnerability.


Should I read the Security Design White Paper?

Yes. Review the 1Password Security Design White Paper before testing, particularly the sections relevant to the component being tested and the section titled Beware of the Leopard.

It explains the intended security model, trust boundaries, cryptographic controls, and known design limitations.


What is the In-Scope Documentation?

The In-Scope Documentation provides practical guidance on:

  • Priority research areas
  • Security boundaries
  • Attacker-story expectations
  • Client, server, and cryptographic controls
  • AI-assisted research
  • Known and intended product behaviors
  • Testing requirements

Review it before submitting a vulnerability.


Can you confirm whether something is in scope, or whether it's intended behavior, before I test?

No. Our security team does not make scope or eligibility determinations outside of a submitted report.

Check the program policy, the In-Scope Documentation, and the Security Design White Paper first. If you believe you've found a real issue but can't yet demonstrate impact, you can submit it as a good-faith flaw hypothesis. See "Scope Questions and Flaw Hypotheses" in the Rules of Engagement.


Are automated scanners allowed?

Yes, but we don't recommend them. Automated scanners are unlikely to uncover the types of vulnerabilities this program is most interested in.

If you do use automation, it must remain below 45 requests per minute. This is approximately 0.75 requests per second, so keep automated traffic below 1 request per second.

Exceeding this limit may result in a temporary block of approximately 24 hours. Blocks caused by exceeding the rate limit will not be lifted early.


Should I identify my testing traffic?

The most important step is signing up for your testing accounts with your @intigriti.me email address. That's our main method for tying test activity to your Intigriti identity.

Where practical, also identify your traffic with a clear User-Agent such as:

Mozilla/5.0 Intigriti-Researcher/<username>

You may also include:

X-Intigriti-Username: <username>

These headers help, but they don't replace using your @intigriti.me email address for your accounts.


Is AI-assisted research allowed?

Yes. AI may be used as part of the research process, but you remain responsible for validating everything you submit.

AI-generated theories must be independently reproduced against the live product and demonstrate a real security impact. See "LLM Prompt / AI-Assisted Research" under In Scope.


What should I do if my account is temporarily blocked?

Contact us via email and provide:

  • Your Intigriti username
  • The affected account (the @intigriti.me email address you signed up with)
  • Approximate testing time
  • Source IP, where relevant
  • The activity you were performing

Removal of a temporary security block cannot be guaranteed.


Is the $1,000,000 Challenge part of this program?

Yes. CTF submissions are made through this program. See "The $1,000,000 Challenge" under In Scope. Only reports containing the actual captured flag qualify.


Can I publicly disclose a vulnerability?

Not without prior written approval from 1Password.

This includes valid, duplicate, resolved, or Not Applicable submissions and public PoCs on platforms such as YouTube, Vimeo, blogs, or social media.

All aboard!
Please log in or sign up on the platform

For obvious reasons we can only allow submissions or applications for our program with a valid Intigriti account.

It will only take 2 minutes to create a new one or even less to log in with an existing account, so don't hesitate and let's get started. We would be thrilled to have you as part of our community.

Researchers
last contributors
leaderboard
Overall stats
submissions received
2
average payout
N/A
accepted submissions
1
total payouts
N/A
Last 90 day response times
avg. time first response
< 1 hour
avg. time to decide
< 1 hour
avg. time to triage
< 1 hour
Activity
11/17
logo
0xdj
created a submission
11/17
1Password updated the confidentiality level to public
11/17
1Password updated the confidentiality level to registered
11/17
1Password updated the confidentiality level to application
11/17
1Password
closed a submission
11/17
1Password
accepted a submission
11/17
logo
1p_ahm2_rschr
created a submission
11/17
1Password Bug Bounty Program
launched