AgentsOpenClaw Supply Chain Crisis 800 Malicious Skills ClawHavocTrend Micro·high signalXBlueskyLinkedInCopy link800+ malicious skills on ClawHub (20% of registry). ClawHavoc planted 1184+ skills distributing Atomic Stealer. 135K exposed instances. 15K vulnerable to RCE.SourceSource pageTrend Micro↳ Follow the threadStack layer / Threat patternClaude Code shipped four releases in five days and the npm 'stable' tag is 27 versions behind 'latest'Creative AI News (verified against npm dist-tags for @anthropic-ai/claude-code)Stack layer / Threat patternSBOM Tools Cover Only the First Two of Four Supply-Chain Propagation StagesarXiv 2609.05380Stack layer / Threat patternCVE-2026-31020: DocsGPT renders user-supplied custom prompts through unsandboxed Jinja, giving unauthenticated RCENVDStack layer / Threat patternTreat an agent's plan as provisional and replace only the suffix runtime evidence invalidates, instead of replanning from scratcharXiv 2609.05019Policy dependency / Threat patternCONTINUITY names "security-context discontinuity" as the failure mode where individually correct agent security controls compose into an insecure systemarXivPolicy dependency / Stack layerMoadim Ships an MIT-Licensed Scheduler That Runs Coding Agents on a Cron Instead of a PromptMoadim, via Hacker News Show HNThreat pattern / ContrastHanding a coding agent false-positive templates doubles its success at writing CodeQL refinements, from 28% to 56-62%arXiv 2609.04535Policy dependency / Stack layerGraphify 0.9.55 ships eight code-graph correctness fixes, including phantom edges fabricated from symbols that own no nodeGitHub