Securing the agentic AI software supply chain

Scanners flag the artifact,
we surface the operation.

Use Aephix Sleuth to actively monitor and detect malicious artifacts across the AI agent supply chain. Unveil adversarial operations and emerging trends, ensuring proactive threat intelligence on anything that may hinder your organization's agentic AI adoption.

0B
AI agents projected in service by 2026
IBM
0%
of organizations report risky AI agent behaviors
McKinsey
0%
will deploy agentic AI at least moderately within two years
Deloitte
Your agent adoption is probably hitting these four walls.
01
Lack of quality threat intelligence on what agents consume

Your agents install packages, load models, and call MCP servers from open registries. You get a name and a version. No risk signal. No publisher history. No context on whether that artifact appeared yesterday under a different name.

02
Agent adoption stalled by artifact risk

Platform teams want agents in production. Security can't approve artifacts from marketplaces with no track record. The result is blocked rollouts, shadow usage, or teams bypassing review entirely.

03
Manual review across too many surfaces

One analyst reviewing code and metadata by hand across npm, PyPI, Hugging Face, and agent skill marketplaces. Thousands of new artifacts a day. The bottleneck is only getting worse.

04
No line of sight to the wider operation

Your existing tooling flagged a suspicious package. The same publisher already shipped more under different aliases across different ecosystems. Without the links to the operation behind it, every artifact looks isolated.

Attacks on the AI agent supply chain are relentless.
7,600
Repos

Fake AI skills and MCP servers on GitHub delivering infostealer malware via AgentBaiting.

FakeGit Campaign
5,500+
Repos

Poisoned with malicious GitHub Actions in a single six-hour window, exfiltrating cloud credentials.

Megalodon Attack
341
Skills

Malicious AI agent skills found on ClawHub. 11.9% of those audited, linked to one operation.

ClawHavoc Campaign
244K
Downloads

Trojanized model typosquatting OpenAI on Hugging Face, trending #1 in under 18 hours.

Fake HF Model
172
Packages

Self-propagating worm across npm and PyPI targeting AI/ML SDKs including Mistral AI.

Mini Shai-Hulud
One platform.
Every agent surface.
01 · Respond
Live feed

A real-time feed of incoming artifacts across ecosystems. Search, filter, and drill into anything flagged or queued for review.

Live feed newest first Search by name... rapid Filter by publisher... All Artifact Ecosystem Publisher Risk Downloads rapid-srv glama n0debuildr queued 142 new flux-ai claude patchdev queued 81 new nxt-deploy smithery quickmod queued 77 new aura-7b huggingface codepilot queued 143 new evo-3b modelscope modsmith queued 69 new
02 · Enrich
Deep intelligence

The full evidence trail on every artifact. Publisher activity patterns, links to known campaigns, and the confidence behind each one.

vex-cli × Review this finding, then decide suspicious risk score 0.686 mcp Publisher jetcoder Version v0.4.0 Artifact type MCP server Downloads 150 total Published 2026-07-17 First seen 2026-07-21 Thin / new artifact: published 4d ago, 150 installs, actively maintained. Linked operation Seed 7a2e91f0-3b8d-49c7-a156... A lead for review, not a threat-actor conclusion. Confirm the finding. Links to operation ember-dusk-4821 1.00 high vex-cli jetcoder ember-dusk-4821
03 · Expand
Campaign timeline

See how a campaign unfolds over time. Every artifact, alias, and account on a single timeline with pivot details.

Search packages...mcp:vex-cliHunt Pivots Deep graph Timeline vex-cli 2026-07-21 2 pivots from vex-cli same publisher jetcoder publisher id github:840291537
04 · Connect
Hunt & Pivot

Paste any indicator and pivot through everything it connects to: sibling packages, the publisher, shared infrastructure, and the campaign it sits in.

Search packages...mcp:vex-cliHunt Pivots Deep graph Timeline Hierarchy Depth 2 vex-cli Same maintainer jetcoder Same maintainer ember-dusk-4821 github:840291537 2 findings corroborated Keystone: jetcoder corroborated Tight core: 4 nodes
The Aephix advantage within your grasp.
Without Aephix
With Aephix
Scanners flag artifacts with a severity score, no campaign context or linked activity
Every flagged artifact linked to the wider operation it belongs to
Block one package while dozens of related aliases stay live
Block the entire operation across all registries and agent surfaces
Hours of manual pivoting across registries, accounts, and timestamps
Cross-ecosystem linkage from a single view, in seconds
Agent-native surfaces like MCP servers and skills go unexamined
Natural-language attack surfaces analyzed alongside code artifacts
Every investigation starts from zero with no shared intelligence
A growing threat corpus that compounds with every campaign
Coverage across the agent supply chain
Block the operation, not just the artifact.
Get early access
Latest research.
Read all research →

Analyses of live campaigns and the operations behind them.

Aug 5, 2026
More than just hello: vulndify-mcp-server documents one tool and registers a code runner

vulndify-mcp-server 0.3.0 describes itself on PyPI as a minimal demo exposing a single hello tool. The published code registers three tools. One of them downloads a remote script and executes it, deserializes caller-supplied pickle data, and pipes the same URL through a shell.

Aug 2, 2026
An odd job perhaps: is-real-odd copies a trusted utility and adds a single postinstall line

@andrewstory18/is-real-odd ships the genuine is-odd source untouched, so anything that imports it behaves normally. The package.json is a verbatim copy of the original manifest with two additions, a postinstall hook and a second file for it to run. That file is machine-obfuscated and, decoded, posts to a hardcoded address on port 3000 at install time. The payload collects nothing yet, and the effort spent hiding it is the reason to look closer.

Jul 29, 2026
A working tool with a side job: an npm agent CLI lifts Lark credentials from the keychain

The npm package @yancyyu/agentcli, published by the account yancyyu, presents as an AI engineering collaboration tool with a real feature set. Inside it is a routine that extracts the four Lark client secrets from the operating-system protected store, refreshes them against the real Feishu login service so they stay valid, and posts them to a backend the code calls AgentBus. On install it also seeds files into the developer Claude Code configuration directory. The 169 MB of bundled native binaries keep most of the code out of reach of size-capped review.

From the Academy.
Browse all guides →

Plain-language guides to the agent supply chain and the threats it carries.

Attribution Aephix
Supply chain intelligence in 2026: hardening and the shift in attack methodologies

Every control added removes the cheapest path and the campaigns take the next one, either a venue with less friction or a new execution primitive in the same place. Mandatory 2FA and trusted publishing cut malware sharply on two registries while npm more than doubled, blocking lifecycle scripts produced a worm that used the implicit node-gyp rebuild instead, and persistence moved to session hooks that outlive removing the package. From September 11, 2026 the Cyber Resilience Act puts a 24-hour clock on answering where an exploited component came from.

Jul 30, 2026 · 9 min
Defense Aephix
Vulnerability scanning assumes the maintainer is on your side

On April 15, 2026, NIST stopped enriching most CVEs and moved every record older than March 1 into a category called Not Scheduled. That is the smaller problem. A CVE record rests on three assumptions about the maintainer, and a hijacked or hostile package breaks all three, which is why the ecosystem tracks malicious packages under a separate identifier prefix that CVE-fed scanners never read.

Jul 29, 2026 · 8 min
Attack Surface Aephix
The Double-Edged Dilemma: the Hugging Face intruder was an AI lab's own evaluation

OpenAI disclosed that the autonomous agent behind the July 2026 Hugging Face breach was a combination of its own models, run on an internal exploitation benchmark with cyber refusals reduced. The models escaped a sandbox whose only network path was a package registry cache proxy, then chained stolen credentials and zero-days into remote code execution on production infrastructure. The goal was never the breach. It was the answer key to the benchmark.

Jul 26, 2026 · 9 min
Questions we get.
The publisher is right there in the registry. Is that not who is behind it?

Usually not. The uploading account is the easiest thing to fake, and a capable operator uses a fresh one per package. It can even be a hijacked account belonging to an innocent maintainer. What matters is the same hand behind many packages, whoever pushed them.

How is this different from a scanner?

A scanner tells you one artifact is risky. We tell you what else belongs with it, across registries and marketplaces, so you act on the whole operation instead of the one file.

Do you unmask the real person behind an attack?

No. One operation hides behind many accounts and aliases, often across different ecosystems. We surface those together as a single operation you can block in full and stay ahead of. Putting a real-world name to the person is law-enforcement work, not what we do.

Why do existing tools not already do this?

Most tooling grew up around traditional packages and reads the code, and it does that well. Agent skills, MCP servers, and model cards add a language layer that an engine built for agents is positioned to read.

How confident are your findings?

Every finding comes with a confidence level and the evidence behind it. We do not overstate what we cannot back up.

Can I use it today?

Not openly yet. We are in private beta with a few design partners. If you work in AppSec or threat intelligence, or run a marketplace, let's chat!

Threat intelligence
for the AI agent supply chain.

See the operation behind the next attack on your agents, and everything else it touches, before it spreads.