New Relic agents

Skip to content What this pays sorona.xyz pays 1.25× the program's reward, up to +$1,000 per finding, plus a complexity bonus of up to $1,000 for especially complex or well-chained bugs. Finding │ Program pays │ Through sorona.xyz P1 (Tier One target) │ $6,000 │ $7,000 P2 (Tier One target) │ $2,500 │ $3,125 P3 (Tier One target) │ $500 │ $625 P4 (Tier One target) │ $250 │ $313 Submit a vulnerability CVE history 0 Last 12 months 3 Lifetime Info disclosure 40 % Code execution 25 % Access control 20 % XSS 10 % Latest advisories Loading the feed Overview New Relic is one of the two or three largest APM vendors in the world, and its entire agent family is Apache-2.0 with no proprietary subtree: the infrastructure agent, the Java, .NET, Node.js, Python, Ruby, Go and PHP agents, the browser agent and the mobile agents are all open source and all actively pushed. Distribution runs through Maven Central, npm, PyPI, RubyGems, NuGet, Packagist, apt and yum, so real install counts run to millions and are invisible in the modest GitHub star counts. Two components carry disproportionate risk. The infrastructure agent runs as root or SYSTEM on every monitored host and executes integration binaries, so a flaw in it is a privileged foothold across a whole fleet at once. The browser agent is a JavaScript snippet injected into customers' public web pages, so a flaw there executes in those customers' end users' browsers and turns one library bug into supply-chain XSS across every site running it. Three CVEs lifetime against that footprint is not reassurance; it is a measure of how little external attention this code has had. New Relic pays through a fully public Bugcrowd engagement at $100 to $6,000, with a flat $6,000 for a P1 against a Tier One target, and commits to aiming for an initial response within 30 days. sorona.xyz adds 1.25× on top of whatever New Relic awards, up to $1,000 more, plus a complexity bonus of up to $1,000 for a well-chained finding. In scope The brief points at github.com/newrelic for the software New Relic distributes and says it may reward security issues found within its agents that could reduce the security of the application the agent is integrated with, assessed against the default configuration. A Tier One Agents list names what counts: Infrastructure, Browser, Android, iOS, Go, Node.js, Ruby, Unity and PHP, with the Java, .NET and Python agents reached through the compatibility documentation The infrastructure agent ( newrelic/infrastructure-agent ), which runs as a privileged process on every monitored Windows and Linux host and executes integrations that shell out to collect data The APM language agents in Java, .NET, Node.js, Python, Ruby, Go and PHP, each of which is loaded into the address space of the instrumented application: the instrumentation hooks, the config and secret handling, the local control surfaces and the data each one captures The browser agent ( newrelic-browser-agent ), a JavaScript snippet injected into customers' production web pages and therefore executing in their end users' browsers The mobile agents for Android, iOS and Unity The agent-to-backend transport: the brief states that traffic between the agents and New Relic backend services may be inspected, and that reports about how the agent connects and transports information are acceptable Kubernetes integrations such as nri-kubernetes , which run inside the cluster they observe. These reach scope through docs.newrelic.com rather than through the Tier One list, so tie the finding to the documented integration Out of scope Every github.com/newrelic repository that is not in the Tier One agent list or on docs.newrelic.com, plus New Relic Example Code, New Relic Experimental and archived repos. The organisation holds hundreds of repositories and is nobody's wildcard Denial of service and distributed denial of service against agents or clients EOL, unmaintained and unsupported agent versions Services hosted by third-party providers, and issues in Salesforce itself on New Relic's Experience Cloud sites Leaked API keys with no demonstrated security impact, for which the brief says plainly that no bounty will be paid Information disclosure through /status or /metrics URLs without a security impact attached to it Reports that refer to a pull request in New Relic's open source projects, which the brief states will not be rewarded Anything filed outside Crowdcontrol: Bugcrowd's terms require submissions to go through it exclusively to be considered for a reward Most-valued findings Remote code execution on a monitored host through an installed agent. The brief names this as a Critical example and it is the P1 band at $6,000. The infrastructure agent is the sharpest version: it runs as root or SYSTEM and executes integration binaries, so one bug is privileged execution across an entire fleet simultaneously. A flaw in the browser agent that lets attacker-controlled content execute in end users' browsers, since that snippet is injected into customers' public pages and a single defect becomes supply-chain XSS across every site running it Sensitive data captured or leaked by an agent that should never have left the instrumented process: request bodies, headers, environment variables, connection strings or credentials appearing in telemetry. Information disclosure is 40% of this code's history and the class the agents' whole job makes most likely. Agent-to-backend transport defects: certificate or hostname validation that can be bypassed, or a response from the backend that an agent trusts far enough to change its own behaviour, which the brief explicitly invites A language agent that weakens the application it instruments: instrumentation that breaks an isolation boundary, deserializes untrusted data, or exposes an agent control surface to code running in the same process Program rules Submit through sorona.xyz first. We validate, coordinate with the program's security team, and add the payout uplift on top of whatever it awards. Findings are judged against the default configuration; an agent that only misbehaves after a configuration change may still be eligible, but say which setting you changed and why it is realistic "Reports that refer to a pull request in our open source projects will not be rewarded" — do not open a fix PR before or alongside the report, which is the exact inverse of how patch-reward programs work Denial of service against agents or clients is out of scope, as are EOL, unmaintained and unsupported agent versions Bugcrowd's Standard Disclosure Terms apply and the brief does not disapply them: submissions must be made exclusively through Crowdcontrol to be considered for a reward, and you assign your testing results to Bugcrowd Only the documented agents count: the github.com/newrelic organisation holds hundreds of repositories and everything not in the Tier One agent list or on docs.newrelic.com is explicitly out Expect a slow clock: New Relic commits only to aiming for an initial response within 30 days Disclosure Coordinated through Bugcrowd under its Standard Disclosure Terms, which make every submission confidential information of the program owner and forbid public disclosure at any time without New Relic's consent. There is no fixed embargo window to count down, so treat publication as New Relic's call, file with Bugcrowd yourself first and route the finding to sorona.xyz afterwards. Safe harbor Sorona will not pursue legal action against you for good-faith research submitted to us in line with this policy. We cannot speak for the upstream project, its vendor, or any third party, and we do not grant authorization on their behalf. You are responsible for understanding and following the laws that apply to you, wherever you are, and for staying within any scope the project has defined. Submit a vulnerability