Over 30 safety vulnerabilities have been disclosed in numerous synthetic intelligence (AI)-powered Built-in Growth Environments (IDEs) that mix immediate injection primitives with reliable options to attain knowledge exfiltration and distant code execution.
The safety shortcomings have been collectively named IDEsaster by safety researcher Ari Marzouk (MaccariTA). They have an effect on well-liked IDEs and extensions reminiscent of Cursor, Windsurf, Kiro.dev, GitHub Copilot, Zed.dev, Roo Code, Junie, and Cline, amongst others. Of those, 24 have been assigned CVE identifiers.
“I believe the truth that a number of common assault chains affected every AI IDE examined is probably the most stunning discovering of this analysis,” Marzouk instructed The Hacker Information.
“All AI IDEs (and coding assistants that combine with them) successfully ignore the bottom software program (IDE) of their risk mannequin. They deal with their options as inherently secure as a result of they have been there for years. Nonetheless, when you add AI brokers that may act autonomously, the identical options will be weaponized into knowledge exfiltration and RCE primitives.”
At its core, these points chain three totally different vectors which can be widespread to AI-driven IDEs –
- Bypass a big language mannequin’s (LLM) guardrails to hijack the context and carry out the attacker’s bidding (aka immediate injection)
- Carry out sure actions with out requiring any person interplay through an AI agent’s auto-approved instrument calls
- Set off an IDE’s reliable options that enable an attacker to interrupt out of the safety boundary to leak delicate knowledge or execute arbitrary instructions
The highlighted points are totally different from prior assault chains which have leveraged immediate injections together with susceptible instruments (or abusing reliable instruments to carry out learn or write actions) to change an AI agent’s configuration to attain code execution or different unintended habits.
What makes IDEsaster notable is that it takes immediate injection primitives and an agent’s instruments, utilizing them to activate reliable options of the IDE to lead to data leakage or command execution.
Context hijacking will be pulled off in myriad methods, together with by means of user-added context references that may take the type of pasted URLs or textual content with hidden characters that aren’t seen to the human eye, however will be parsed by the LLM. Alternatively, the context will be polluted by utilizing a Mannequin Context Protocol (MCP) server by means of instrument poisoning or rug pulls, or when a reliable MCP server parses attacker-controlled enter from an exterior supply.
Among the recognized assaults made attainable by the brand new exploit chain is as follows –
- CVE-2025-49150 (Cursor), CVE-2025-53097 (Roo Code), CVE-2025-58335 (JetBrains Junie), GitHub Copilot (no CVE), Kiro.dev (no CVE), and Claude Code (addressed with a safety warning) – Utilizing a immediate injection to learn a delicate file utilizing both a reliable (“read_file”) or susceptible instrument (“search_files” or “search_project”) and writing a JSON file through a reliable instrument (“write_file” or “edit_file)) with a distant JSON schema hosted on an attacker-controlled area, inflicting the info to be leaked when the IDE makes a GET request
- CVE-2025-53773 (GitHub Copilot), CVE-2025-54130 (Cursor), CVE-2025-53536 (Roo Code), CVE-2025-55012 (Zed.dev), and Claude Code (addressed with a safety warning) – Utilizing a immediate injection to edit IDE settings recordsdata (“.vscode/settings.json” or “.concept/workspace.xml”) to attain code execution by setting “php.validate.executablePath” or “PATH_TO_GIT” to the trail of an executable file containing malicious code
- CVE-2025-64660 (GitHub Copilot), CVE-2025-61590 (Cursor), and CVE-2025-58372 (Roo Code) – Utilizing a immediate injection to edit workspace configuration recordsdata (*.code-workspace) and override multi-root workspace settings to attain code execution
It is value noting that the final two examples hinge on an AI agent being configured to auto-approve file writes, which subsequently permits an attacker with the flexibility to affect prompts to trigger malicious workspace settings to be written. However on condition that this habits is auto-approved by default for in-workspace recordsdata, it results in arbitrary code execution with none person interplay or the necessity to reopen the workspace.
With immediate injections and jailbreaks appearing as step one for the assault chain, Marzouk affords the next suggestions –
- Solely use AI IDEs (and AI brokers) with trusted tasks and recordsdata. Malicious rule recordsdata, directions hidden inside supply code or different recordsdata (README), and even file names can develop into immediate injection vectors.
- Solely connect with trusted MCP servers and constantly monitor these servers for modifications (even a trusted server will be breached). Evaluate and perceive the info movement of MCP instruments (e.g., a reliable MCP instrument may pull data from attacker managed supply, reminiscent of a GitHub PR)
- Manually evaluation sources you add (reminiscent of through URLs) for hidden directions (feedback in HTML / css-hidden textual content / invisible unicode characters, and so forth.)
Builders of AI brokers and AI IDEs are suggested to use the precept of least privilege to LLM instruments, decrease immediate injection vectors, harden the system immediate, use sandboxing to run instructions, carry out safety testing for path traversal, data leakage, and command injection.
The disclosure coincides with the invention of a number of vulnerabilities in AI coding instruments that would have a variety of impacts –
- A command injection flaw in OpenAI Codex CLI (CVE-2025-61260) that takes benefit of the truth that this system implicitly trusts instructions configured through MCP server entries and executes them at startup with out in search of a person’s permission. This might result in arbitrary command execution when a malicious actor can tamper with the repository’s “.env” and “./.codex/config.toml” recordsdata.
- An oblique immediate injection in Google Antigravity utilizing a poisoned internet supply that can be utilized to govern Gemini into harvesting credentials and delicate code from a person’s IDE and exfiltrating the knowledge utilizing a browser subagent to browse to a malicious website.
- A number of vulnerabilities in Google Antigravity that would lead to knowledge exfiltration and distant command execution through oblique immediate injections, in addition to leverage a malicious trusted workspace to embed a persistent backdoor to execute arbitrary code each time the applying is launched sooner or later.
- A brand new class of vulnerability named PromptPwnd that targets AI brokers related to susceptible GitHub Actions (or GitLab CI/CD pipelines) with immediate injections to trick them into executing built-in privileged instruments that result in data leak or code execution.
As agentic AI instruments have gotten more and more well-liked in enterprise environments, these findings exhibit how AI instruments increase the assault floor of growth machines, usually by leveraging an LLM’s incapability to differentiate between directions supplied by a person to finish a process and content material that it could ingest from an exterior supply, which, in flip, can include an embedded malicious immediate.
“Any repository utilizing AI for challenge triage, PR labeling, code strategies, or automated replies is susceptible to immediate injection, command injection, secret exfiltration, repository compromise and upstream provide chain compromise,” Aikido researcher Rein Daelman mentioned.
Marzouk additionally mentioned the discoveries emphasised the significance of “Safe for AI,” which is a brand new paradigm that has been coined by the researcher to sort out safety challenges launched by AI options, thereby making certain that merchandise will not be solely safe by default and safe by design, however are additionally conceived holding in thoughts how AI parts will be abused over time.
“That is one other instance of why the ‘Safe for AI’ precept is required,” Marzouk mentioned. “Connecting AI brokers to present functions (in my case IDE, of their case GitHub Actions) creates new rising dangers.”
