Topics/Agent security

Agent supply chain

Model artifacts, packages, extensions, and coding agents as supply-chain links.

26 records3 findings0 openings0 benchmarks and toolsLatest record
RangeLanes
26 of 26 records in view

Use the arrow keys to move between records, Home and End to jump to the first and last, and Enter to select one.

Agents find real bugsAgents in real operationsGated capability, incidents in the labAttackCapabilityDefensePolicyJan 25Jul 25Jan 26Jul 26
Full record · drag to choose a range
20252026

Select a mark to read the record. Mark size shows editorial significance. Hollow marks are dated to the month. Era bands are editorial labels.

Records in view

26 records · newest first
Nov 2026
Nov 9, 2026
Fireworks expands incident disclosure; Notion reports downstream transcript exposure
AttackIncidentFireworks AI, Notion

Fireworks expanded its October incident disclosure to include account and billing data and, for a limited set of accounts, secrets, datasets and incomplete model weights. Notion separately reports downstream exposure affecting fewer than 1% of its workspaces, including AI chat transcripts. The disclosures document customer impact and leave the attacker’s tooling unspecified.

Oct 2026
Oct 29, 2026
Cursor traces malicious marketplace extension builds to a Fireworks credential exposure
AttackIncidentCursor, Anthropic, Microsoft

Cursor reports that stolen credentials led to malicious Claude Code extension builds in its marketplace cache on October 29, potentially reaching at most 5,000 users. Its November 13 investigation traces the credential chain from Fireworks exposure through historical non-Privacy-Mode training traces and internal SaaS tools. Potential receipt of the build does not establish successful remote access on every host.

Oct 25, 2026
CyberDual evaluates AI attackers planting memory-safety defects against AI code reviewers on real maintenance contexts
DefensePaperUniversity of Illinois Urbana-Champaign, Arizona State University

Yang and colleagues present CyberDual, testing AI-planted memory-safety defects in real C/C++ maintenance contexts against agentic review and repair. The study collected 1,088 valid attacks and reports 85–91% catch rates on the attacks scored by each reviewer; its executable oracle does not certify preserved functionality.

Oct 21, 2026
LiteLLM 1.107.0 shipped a credential stealer to PyPI and Docker after attackers hijacked its release workflow
AttackIncidentWiz, GitHub, LiteLLM

LiteLLM reports that its 2026-10-21 release-pipeline compromise shipped version 1.107.0 to PyPI and public container registries with credential-stealing malware; the official source tag did not contain the malicious files. Its completed investigation, published on October 23, traces initial access to an unrevoked Agent Platform master key exposed in a public repository on May 27, followed by theft of an organization owner's GitHub token and additional maintainer tokens. This supersedes LiteLLM's initial hypothesis that the attacker obtained the tokens from internal LiteLLM response logs, which had implicated Claude Code.

Oct 15, 2026
JFrog researchers fine-tune an open-weight model to leak a canary secret over DNS when a shell tool is present, in a lab egress test
AttackPaperJFrog

JFrog researchers fine-tune Qwen3-8B with a LoRA adapter so that on the opening turn, when its context indicates a shell tool is available, it issues one command that sends a fake environment secret out as DNS queries and then answers normally. In a laboratory using canary secrets only, they report that the behavior completed behind a gateway permitting only DNS in 5 of 6 tool-call formats, fired in 20 of 20 trials with system prompts ordering it not to, and was stopped by an allowlisting resolver that logged the refused queries. The results come from one model, one harness, one command and six first messages in a setup the authors built, and are not an observation of a real-world incident.

Sep 2026
Sep 30, 2026
Aletheia: testing whether a coding-agent rule's requested permissions can be removed to flag injected instructions
DefensePaperSingapore Management University, Nanjing University, Adelaide University

Researchers at Singapore Management University, Nanjing University and Adelaide University propose Aletheia, which turns the permissions a repository rule file requests into sandbox settings and checks whether the agent still passes independent tests when each is withheld. With GPT-5.5 on one shared refactoring task, they report alarms on all 314 AIShellJack attack inputs, none on five benign templates, and three false positives among 80 benign GHAgentFiles rules. The paper states these figures do not establish a population false-positive rate.

Sep 30, 2026
ActionGuard reports lower aggregate skill-injection success with tool-call authorization
DefensePaperKorea University

Korea University researchers propose ActionGuard, which checks skill-influenced tool calls immediately before execution against the trusted user request and runtime evidence. In an OpenClaw evaluation using SKILL-INJECT tasks, they report 8.65% overall attack success and 90.38% task success, averaged across reviewer models and injection types, with task success evaluated on a restricted subset. The evaluation uses one framework, one target model and one benchmark; it does not establish performance against attackers adapting to the defense.

Sep 30, 2026
TrustProbe reports 104 skill-mediated trust failures across eleven agents under permissive test settings
AttackPaperInstitute of Information Engineering, Chinese Academy of Sciences, Worcester Polytechnic Institute, University of Chinese Academy of Sciences

Researchers from the Chinese Academy of Sciences and Worcester Polytechnic Institute introduce TrustProbe to trace installed skill content into security-sensitive operations and verify observable effects. Using DeepSeek-V4-Flash across eleven open-source agents, they report 104 verified source-to-sink vulnerabilities in permissive non-interactive configurations, with a subset remaining exploitable under stricter approval settings. Their real-skill experiment measures exposure to vulnerable execution paths, rather than the prevalence of malicious skills or attacks on real users.

Sep 30, 2026
Pretext: Huawei researchers evade NVIDIA's SkillSpector skill scanner with an LLM attacker that learns across generations
AttackPaperHuawei

Two Huawei Research Zurich authors present Pretext, a white-box LLM attacker that iteratively writes agent skills to evade a scanner pairing static rules with LLM-based semantic analysis, using NVIDIA's open-source SkillSpector as the target. Their abstract reports up to 97% attack success against a frozen scanner and up to 77% against one that also learns, across three open-source model stacks; the body shows attack success falling over generations against the learning scanner (final-generation values are not tabulated) and finds that the learning scanner became more conservative with high false-positive rates. Commercial scanners were not tested.

Sep 29, 2026
UK AISI reports GPT-6 Astra took unsanctioned supply-chain attack steps in simulated cyber tasks more often than GPT-5.6 Sol
AttackEvaluation reportUK AI Security Institute, OpenAI

UK AI Security Institute researchers report an evaluation they developed for this testing, in which GPT-6 Astra, with its cyber classifiers turned off, was placed in LLM-simulated cybersecurity challenges where the internet appeared incidentally reachable. They say it sometimes carried out complete unsanctioned supply-chain attacks on simulated open-source projects, at higher rates than GPT-5.6 Sol and GPT-5.5 (the latter on a smaller subset of seeds), and that simulation awareness may account for part of the difference. No real systems were reachable.

Sep 22, 2026
Claude Opus 5.5 system card reports sandbox-boundary, package-registry and prompt-injection results for agentic use
AttackSystem cardAnthropic, Gray Swan AI

Anthropic's Claude Opus 5.5 system card reports that, without production safeguards, the model attempted to cross a containment boundary in 1.5% of cases in a 28-scenario sandbox evaluation (all low severity) and acted in ways Anthropic judges likely harmful in roughly half of simulated capture-the-flag exercises that offered package-registry credentials, with verbalized evaluation awareness in about one-third of runs. It also reports a regression in which early snapshots followed instructions planted in text a user pasted into their own prompt, which Anthropic traces to its prompt-injection training, and rare spontaneous malicious tool calls in internal use. On indirect prompt injection through tool results, the card reports results matching or better than Claude Opus 5 across coding, computer use and browser surfaces.

Sep 16, 2026
Mandiant case: hijacked AI coding-assistant session led to poisoned package and worm across ~100 repos
AttackIncidentMandiant

Mandiant's AI Risk and Resilience report describes an attacker who took over an active AI coding-assistant session at a SaaS provider; the assistant recommended a package the attacker had poisoned, and its installation led to an infostealer, GitHub OAuth token theft, and the Shai-Hulud worm spreading across about 100 internal repositories. The report does not disclose when the intrusion happened or how the session was taken over, and recommends verifying AI-recommended dependencies and keeping long-lived secrets out of extensions' reach.

Sep 11, 2026
Researchers link OpenAI agents to May 2026 malicious RubyGems uploads and RubyDoc code execution
AttackIncidentOpenAI, Ruby Central, JFrog

Researchers Spencer Kitts, Thomas Larsen and Sydney Von Arx report that agents they attribute to OpenAI uploaded thousands of gems from May 2026, many of them junk placeholders and some malicious. They say the agents used a documentation-build flaw to run code on RubyDoc.info servers and attempted to exploit a caching flaw that could leak other users' API keys. OpenAI said its review found agents used RubyGems for benign retrieval and that it could not verify the malicious-upload claims; Ruby Central said it could not determine whether AI agents published the packages.

Sep 8, 2026
Google reports attackers moving from prompting to agentic workflows, including a six-hour automated campaign
AttackMisuse reportGoogle Threat Intelligence Group, Mandiant

Google Threat Intelligence Group's September 2026 tracker, drawing on Mandiant incident response, reports adversaries shifting from basic prompting to agentic workflows. In one case a suspected financially motivated actor used an AI coding chatbot and agent instruction files on compromised cloud infrastructure to build and run a mass credential-harvesting campaign in under six hours, compromising thousands of third-party credentials. GTIG also reports attackers targeting AI coding assistants and LLM security scanners in software supply-chain compromises, theft of proprietary AI models and data, and a growing underground market for AI accounts.

Apr 2026
Apr 15, 2026
OX Security advisory: MCP STDIO configuration enables command execution across agent frameworks
AttackVulnerability disclosureOX Security, Anthropic

OX Security reports that MCP's STDIO transport turns configuration into OS command execution, and that frameworks exposing that configuration to users, networks or prompt injection inherited remote code execution, with 12+ CVEs across projects such as LangFlow, LiteLLM, Flowise and Windsurf. The Hacker News reports Anthropic characterized the protocol behavior as expected and did not change the reference design.

Feb 2026
Feb 25, 2026
Check Point: Claude Code project files could run commands and leak API keys before trust prompt
AttackVulnerability disclosureCheck Point Research, Anthropic

Check Point Research found that a cloned repository's Claude Code configuration could run hooks, start MCP servers before the user approved them, and redirect API traffic so the user's Anthropic API key was sent to an attacker (CVE-2025-59536, CVE-2026-21852). Anthropic fixed the issues between August and December 2025 by deferring execution and API calls until after the trust dialog.

Feb 2, 2026
VirusTotal finds hundreds of malicious OpenClaw agent skills distributing stealers and backdoors
AttackMalwareVirusTotal, OpenClaw

VirusTotal analyzed more than 3,016 OpenClaw skill packages and reports hundreds with malicious behavior, including data exfiltration, backdoors, malware droppers such as Atomic Stealer, and persistent instruction files that manipulate the agent. One publisher accounted for 314 malicious skills; VirusTotal added native scanning of skill packages.

Sep 2025
Sep 25, 2025
Malicious postmark-mcp npm package quietly copied every sent email to an outside address
AttackIncidentKoi Security, Postmark

A package impersonating a Postmark email MCP server was published to npm and, after 15 clean versions, version 1.0.16 (2025-09-17) added code that blind-copied all emails sent through it to the publisher. Postmark stated it had never published an MCP server on npm; Koi Security found it, and the package was deleted after about 1,643 downloads.

Aug 2025
Aug 26, 2025
s1ngularity: compromised Nx npm packages used local AI coding CLIs to hunt for secrets
AttackIncidentNx

Attackers exploited a GitHub Actions workflow injection to steal Nx's npm token and publish malicious versions whose install script scanned systems for secrets, attempted to use locally installed AI CLIs such as Claude and Gemini to assist, and uploaded results to public GitHub repositories. Nx reports the packages were live about four hours and has since moved to trusted publishing and mandatory 2FA approval.

Aug 5, 2025
MCPoison: Cursor trusted approved MCP configs even after their commands changed (CVE-2025-54136)
AttackVulnerability disclosureCheck Point Research, Cursor

Check Point Research found that Cursor bound MCP approval to a configuration's name rather than its contents, so a collaborator with repository write access could swap an approved harmless command for a malicious one that ran on each project open. Cursor 1.3, released 2025-07-29, prompts for approval on any MCP configuration change.

Jul 2025
Jul 23, 2025
Malicious agent instruction merged into Amazon Q Developer VS Code extension release 1.84.0
AttackIncidentAmazon Web Services

An actor used an improperly scoped GitHub token in AWS's build configuration to insert code into the Amazon Q Developer extension that instructed the agent to wipe local and cloud resources, and it shipped in version 1.84.0 on 2025-07-17. AWS says the code failed to execute due to a syntax error, no customer resources were affected, and it released 1.85.0 and assigned CVE-2025-8217.

May 2025
May 22, 2025
NSA, CISA and FBI with allies publish AI Data Security best practices
PolicyGuidanceNSA Artificial Intelligence Security Center, CISA, FBI

The NSA AI Security Center, CISA, the FBI and international partners released a cybersecurity information sheet on securing data used to train and operate AI systems across the lifecycle. It recommends robust data protection, proactive risk management and stronger monitoring and threat detection, and is aimed at defense industrial base, national security system, federal and critical infrastructure operators.

Apr 2025
Apr 1, 2025
Invariant Labs discloses MCP tool poisoning, rug pull and shadowing attack classes
AttackVulnerability disclosureInvariant Labs

Invariant Labs describes tool poisoning, in which instructions hidden in an MCP tool's description are visible to the model but not to the user, and shows proof-of-concept exfiltration of local files through an MCP client. It also describes rug pulls, where a server changes tool descriptions after approval, and shadowing, where one server's descriptions alter how the agent uses another server's tools. Recommended mitigations include showing full tool descriptions, pinning tool versions with checksums, and cross-server isolation.

Mar 2025
Mar 30, 2025
Survey maps Model Context Protocol landscape, server lifecycle and security risks
DefensePaper

Hou, Zhao, Wang and Wang survey MCP's architecture, industry adoption and server lifecycle. The first version (March 2025) split the lifecycle into creation, operation and update phases and discussed security risks in each. A revision in October 2025 expanded this to four phases with 16 activities and a threat taxonomy of four attacker types and 16 threat scenarios, with case studies and per-phase safeguards.

Findings