Paid bug bounty / Launched May 25, 2025
Find the boundary.
Help us strengthen it.
Distributed Systems, Inc. pays for original, reproducible security vulnerabilities in our software. Five graduated tiers reward demonstrated impact, from $20 to $750 in USD cash.
Five graduated reward tiers
Impact determines the reward.
These are cash awards, not product credits. We assess exploitability, required privileges, affected data, and the reach of the underlying flaw. A vulnerability may fall below a tier when its demonstrated impact is narrower than the example.
- Tier 01
Low
A reproducible security weakness with limited exposure or a narrow impact, beyond a purely cosmetic or best-practice observation.
$20USD cash - Tier 02
Moderate
A limited disclosure or permission bypass affecting a researcher-controlled account or a restricted set of non-sensitive resources.
$75USD cash - Tier 03
High
A substantial access-control failure, account takeover, or unauthorized action with demonstrated impact in an approved test environment.
$200USD cash - Tier 04
Critical
A demonstrated cross-tenant boundary failure, sensitive-secret exposure, or arbitrary execution across a significant trust boundary in an approved environment.
$500USD cash - Tier 05
Exceptional critical
A reliable exploit or chain with broad impact across the agent platform, such as privileged execution or a systemic containment failure, demonstrated safely in an approved environment.
$750USD cash
One award per distinct root cause, paid to the first complete, eligible report. A chain is assessed as one finding based on its combined impact. An isolated reproduction is sufficient; do not access other people’s data or expand an exploit to prove a larger impact.
Scope & authorization
Our software.
Your isolated test.
Open scope: local, isolated testing of company-owned source that we publish and identify as DSCO, Chimera, Tool Management API, or GraphSub, including their authentication, authorization, routing, tool-execution and agent-containment code. Use source linked from our official product pages or the DSCO repository. Only code we own or maintain is included.
Non-disruptive checks of the public distributed.systems marketing website are also in scope. Customer data, other tenants, production accounts and backend infrastructure are outside this open authorization.
Ask before testing live services. Active testing of production APIs, authentication, model gateways or execution services requires Arthur’s written approval naming the exact assets, test accounts, techniques and dates. A product listing, subdomain or shared hosting address does not provide permission to test it.
Third-party model providers, identity providers, cloud platforms, customer systems and personal accounts are excluded. Historical Stickerfacet reports are documented separately; its former deployment is not an active bounty target.
Submit a finding
Make it reproducible.
Keep it private.
Send reports to arthur@distributed.systems, with “DSCO bug bounty” in the subject. Include:
- Affected product, repository, revision or approved asset, and your authorization reference for live tests.
- Clear steps to reproduce using synthetic data and your own test accounts.
- Expected behavior, observed behavior, impact and minimal sanitized evidence.
- A suggested fix if available, and how to reach you for follow-up.
Do not email credentials, customer records or full exploit dumps. Ask for a secure transfer method if sensitive evidence is necessary. We aim to acknowledge reports within three business days and provide an initial triage decision within ten business days.
Eligibility & payment
Clear terms.
A documented decision.
A report must identify a previously unknown, reproducible security vulnerability in scope, follow these rules, and demonstrate meaningful impact. Eligible validated reports receive the published reward for the assigned tier. We explain severity and duplicate decisions; request a review by replying with additional evidence.
Informational observations, automated scanner output without validation, unsupported version warnings, cosmetic issues, social engineering, spam, denial of service, brute force, and reports requiring prohibited access are not reward-eligible. Current employees and contractors are not eligible for awards arising from their assigned work.
After validation, eligibility checks and agreement on a lawful payment method, we pay the reward in USD within 30 calendar days. Required payment and tax details are collected privately after acceptance. No fee, purchase or product-credit redemption is required.
Use the minimum necessary proof, stop if you encounter another person’s data, and report the accidental exposure without retaining it. Do not modify or delete data, persist access, disrupt service, or disclose the finding publicly before coordinated remediation. Recognition is optional; anonymity is welcome.
We authorize good-faith research within the scope and conditions above and will not pursue legal action against researchers for that authorized work. This authorization covers our assets only and cannot authorize access to third-party systems.
Program changes will be dated on this page. Reports submitted while the program is active are evaluated under the published terms in force when received. Effective and launch date: May 25, 2025.
Program history
A new paid program.
Earlier security work.
This paid program launched May 25, 2025. The May 2025 Stickerfacet reports document earlier vulnerability handling. View our security function and assessment evidence for the broader record.