--- title: Claude Tag lands in Slack, AI agent skill marketplace... description: A roundup of major AI and security developments: Anthropic launched Claude Tag, embedding Claude as a persistent ambient AI teammate inside Slack channels for... canonical: https://daily.dev/posts/claude-tag-lands-in-slack-ai-agent-skill-marketplace-hit-by-supply-chain-attack-rr6lfdngu twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Claude Tag lands in Slack, AI agent skill marketplace hit by supply-chain attack | daily.dev og:description: A roundup of major AI and security developments: Anthropic launched Claude Tag, embedding Claude as a persistent ambient AI teammate inside Slack channels for... og:url: https://daily.dev/posts/claude-tag-lands-in-slack-ai-agent-skill-marketplace-hit-by-supply-chain-attack-rr6lfdngu og:image: https://api.daily.dev/og/posts/rR6lfDnGU.png og:image:alt: Claude Tag lands in Slack, AI agent skill marketplace hit by supply-chain attack og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Claude Tag lands in Slack, AI agent skill marketplace hit by supply-chain attack **[Agentic Digest](https://daily.dev/sources/agents_digest)** · 5 min read · 1 upvotes · 0 comments ## Summary A roundup of major AI and security developments: Anthropic launched Claude Tag, embedding Claude as a persistent ambient AI teammate inside Slack channels for Enterprise and Team customers. A security firm exposed a critical supply-chain flaw in AI agent skill marketplaces — a fake skill bypassed all major scanners and reached 26,000 agents by swapping a benign URL for a malicious payload post-approval. Z.ai's GLM-5.2 is emerging as the strongest open-weights model, scoring near Claude Opus 4.7 and GPT-5.5 with a 1M token context window, though with notable caveats around benchmark optimization and distillation. The Miasma worm compromised 89+ npm packages in June 2026, using stolen credentials and novel techniques including valid SLSA provenance attestations and persistence mechanisms targeting Claude Code, Cursor, and Gemini CLI. Additional items cover a $6.3B SpaceX compute deal, four CVEs in Dify, Menlo Ventures' $3B fund, an AI hiring bias lawsuit against Workday, and DigitalOcean's LLM cost-routing tool. ## Content **TLDR:** Anthropic launched Claude Tag in research preview, embedding Claude as a persistent, ambient teammate inside Slack channels for Enterprise and Team customers. A security firm demonstrated a fake agent skill that bypassed every major scanner and reportedly reached 26,000 agents by swapping a benign URL for a malicious payload post-approval. GLM-5.2 from Z.ai is drawing serious attention as the strongest open-weights model currently available, scoring close to Claude Opus 4.7 and GPT-5.5. The Miasma worm compromised 89+ npm packages in June 2026, with a novel persistence mechanism specifically targeting AI coding assistants including Claude Code, Cursor, and Gemini CLI. --- ## Anthropic launches Claude Tag: persistent, ambient AI in Slack channels Claude Tag gives Claude its own identity and dedicated service accounts inside Slack, replacing the old on-demand integration with something that actually watches channels and intervenes proactively — flagging forgotten threads, surfacing relevant context, and working through multi-step tasks asynchronously. Admins scope each Claude identity to specific channels and tools, which addresses the obvious multiplayer security concern. It puts Anthropic directly against Microsoft Copilot Work IQ, Glean, and GitHub Copilot in Teams in the race to own organizational context. The privacy implications of an always-listening workplace AI are real and worth thinking through before rolling it out. [Read more](https://daily.dev/feed-by-ids?id=Be6JaCH80&id=6qZQ7axS0&id=IRmfrXPRI&id=0XENBQsPp) ## Fake agent skill bypassed every scanner, reportedly hit 26,000 agents Security firm AIR published a malicious skill to a popular marketplace, passed scanners from Cisco and NVIDIA, then swapped the URL it pointed to from legitimate Google Stitch docs to a script-download page after widespread installation. Trail of Bits independently confirmed the same bypass technique. The structural flaw is that scanners analyze a package once at submission time, while the external URL can be changed later. NVIDIA's own SkillSpector scanner — released separately — detects 65 vulnerability patterns across 16 categories and found that 26.1% of agent skills contain vulnerabilities and 5.2% show likely malicious intent. Treat skills as software: pin versions, vet external links, route installs through controlled sources. [Read more](https://daily.dev/feed-by-ids?id=bDCAQioYV&id=jpVAd71U5) ## GLM-5.2 is the strongest open-weights model right now, with caveats Z.ai's GLM-5.2 scores close to Claude Opus 4.7-4.8 and GPT-5.5 on benchmarks, with a 1 million token context window and strong coding and long-context performance. The caveats are real: it appears heavily distilled from Claude (causing identity confusion), lacks native vision, is expensive for an open model, and shows signs of benchmark optimization. Vercel CEO Guillermo Rauch and former Meta/Google VP Mat Velloso have both praised it publicly. Right pick for hard tasks requiring an open model; don't overhype it relative to top closed models. [Read more](https://daily.dev/feed-by-ids?id=T0V8gCU3d&id=Q9fazdQR4) ## Miasma worm targets AI coding assistants via stolen developer credentials The Miasma worm compromised 89+ npm packages across three waves in June 2026, hitting Red Hat, Vapi.ai, and Microsoft Azure repositories. The root cause was a stolen developer credential that sat in underground markets for seven weeks before use. What's new: malicious packages carried valid SLSA Build Level 3 provenance attestations, a 'Phantom Gyp' technique bypassed install-script monitoring, and a persistence mechanism specifically drops config file overrides into Claude Code, Cursor, Gemini CLI, and VS Code. The full attack toolchain is now public, permanently lowering the bar for copycats. Recommended response: dark web credential monitoring with automated rotation, human-gated npm publishing, and treating developer environments as control-plane infrastructure. [Read more](https://daily.dev/posts/tx6DkxVF3) --- ## Also notable - **SpaceX compute landlord: Reflection AI signs $6.3B deal for GB300 access through 2029:** Reflection AI will pay $150M/month to rent Nvidia GB300 chips at SpaceX's Colossus 2 data center, giving SpaceX over $80B in committed compute revenue through 2029 alongside existing contracts with Anthropic (~$1.25B/month) and Google (~$920M/month) — and Reflection is now raising again at a $25B valuation despite not yet shipping a public frontier model. [Read more](https://daily.dev/posts/fRwyD0nKg) - **DifyTap: four CVEs in Dify including CVSS 9.4 path traversal, patched in 1.14.2:** Zafran researchers found a tracing hijack (CVE-2026-41947, CVSS 9.1) enabling silent chat history interception, a Plugin Daemon path traversal (CVE-2026-41948, CVSS 9.4) exposing internal APIs to unauthenticated requests, and two UUID-based cross-tenant document access bugs — three of four are patched in Dify 1.14.2. [Read more](https://daily.dev/posts/bGIqwgadW) - **Menlo Ventures raises $3B fund after its $750M Anthropic Series D bet grew to ~$14B:** Menlo's Anthropic stake, structured as a $500M SPV alongside a $250M direct investment, is now worth approximately $14B, and the firm's Anthology co-fund with Anthropic has deployed ~$250M across 60+ companies with exits including Graphite (acquired by Cursor) and Astrix Security (acquired by Cisco). [Read more](https://daily.dev/posts/Xs2AscPLC) - **Workday must face class action over AI hiring bias — first case targeting algorithmic screening vendors:** A US federal judge ruled Workday cannot escape Mobley v. Workday, the first lawsuit to broadly target AI screening tool vendors rather than individual employers, allowing discrimination claims related to race, age, and disability to proceed — a ruling that could reshape liability for algorithmic hiring platforms across the industry. [Read more](https://daily.dev/posts/SqN7w1DtC) - **DigitalOcean Inference Router cuts LLM costs 39.6% at scale with no code changes:** Routing 700K classify, 250K Q&A, and 50K reasoning requests/month through DigitalOcean's Inference Router costs $2,850/month versus $4,716 for a hardcoded Claude Sonnet 4.6 baseline, with session pinning via X-Model-Affinity preserving KV-cache warmth across multi-turn conversations. [Read more](https://daily.dev/posts/JTiIgUkdg) ## Similar posts on daily.dev - [Claude Tag Security: Governing AI Identities in Slack](https://daily.dev/posts/claude-tag-security-governing-ai-identities-in-slack-niiscvjov) · Security Boulevard · 0 upvotes · 0 comments - [Anthropic’s Claude Tag aims to turn workplace AI from a personal assistant into a teammate](https://daily.dev/posts/anthropic-s-claude-tag-aims-to-turn-workplace-ai-from-a-personal-assistant-into-a-teammate-jbzs4iqhn) · InfoWorld · 1 upvotes · 0 comments - [Anthropic puts Claude inside Slack as a tagable co-worker](https://daily.dev/posts/anthropic-puts-claude-inside-slack-as-a-tagable-co-worker-9ncf1jwhw) · TechCentral · 0 upvotes · 0 comments --- [View this post on daily.dev](https://daily.dev/posts/claude-tag-lands-in-slack-ai-agent-skill-marketplace-hit-by-supply-chain-attack-rr6lfdngu) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/claude-tag-lands-in-slack-ai-agent-skill-marketplace-hit-by-supply-chain-attack-rr6lfdngu","headline":"Claude Tag lands in Slack, AI agent skill marketplace hit by supply-chain attack","text":"A roundup of major AI and security developments: Anthropic launched Claude Tag, embedding Claude as a persistent ambient AI teammate inside Slack channels for Enterprise and Team customers. A security firm exposed a critical supply-chain flaw in AI agent skill marketplaces — a fake skill bypassed all major scanners and reached 26,000 agents by swapping a benign URL for a malicious payload post-approval. Z.ai's GLM-5.2 is emerging as the strongest open-weights model, scoring near Claude Opus 4.7 and GPT-5.5 with a 1M token context window, though with notable caveats around benchmark optimization and distillation. The Miasma worm compromised 89+ npm packages in June 2026, using stolen credentials and novel techniques including valid SLSA provenance attestations and persistence mechanisms targeting Claude Code, Cursor, and Gemini CLI. Additional items cover a $6.3B SpaceX compute deal, four CVEs in Dify, Menlo Ventures' $3B fund, an AI hiring bias lawsuit against Workday, and DigitalOcean's LLM cost-routing tool.","url":"https://daily.dev/posts/claude-tag-lands-in-slack-ai-agent-skill-marketplace-hit-by-supply-chain-attack-rr6lfdngu","datePublished":"2026-06-24T04:18:14.330Z","dateModified":"2026-06-24T04:18:37.519Z","author":{"@type":"Organization","name":"Agentic Digest","logo":"https://media.daily.dev/image/upload/s--V91DY4ls--/f_auto,q_auto/v1772617267/logos/agents_digest","url":"https://daily.dev/sources/agents_digest"},"interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/sources/agents_digest","name":"Agentic Digest"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Agentic Digest","item":"https://daily.dev/sources/agents_digest"},{"@type":"ListItem","position":3,"name":"Claude Tag lands in Slack, AI agent skill marketplace hit by supply-chain attack"}]} ```