Blog
Insights, updates, and deep dives on AI agents, decentralized standards, and the future of HOL.
HOL Guard 3.0: Rust authority, extension-first controls, honest harness contracts
HOL Guard 3.0 puts PreToolUse on a version-matched Rust path with no Python semantic fallback, makes Local Extensions the capability vocabulary for managed-restrictive Control Sets, and keeps sync paused from meaning unprotected. Pin tip 3.0.113; do not pin yanked 3.0.0.
BREAKING: CVE-2026-76578 lets an unauthenticated LDAP client become a FreeIPA admin
How to fix CVE-2026-76578: firewall FreeIPA LDAP ports 389/636 and disable anonymous binds until the IPA/IdM package with the hardened OTP ACI ships
HOL Guard
BREAKING: CVE-2026-86259 lets unauth OpenMAIC callers pull cloud credentials via SSRF
How to fix CVE-2026-86259: upgrade OpenMAIC to 1.0.1
HOL Guard
BREAKING: CVE-2026-67276 and MikroTrick can take over MikroTik RouterOS with SSH exposed
How to fix CVE-2026-67276: upgrade RouterOS to 7.24.2, 7.23.4, or 6.49.21
HOL Guard
BREAKING: CVE-2026-0799 lets crafted BPF filters walk libpcap process memory
How to fix CVE-2026-0799: upgrade libpcap to 1.10.7 (covers six sibling CVEs)
HOL Guard
CVE-2026-85787: AWS postgres MCP read-only denylist missed set_config()
How to fix CVE-2026-85787: upgrade awslabs.postgres-mcp-server to 1.1.7 or newer
HOL Guard
BREAKING: CVE-2026-9317 lets anyone who can reach your Nango runner run code
How to fix CVE-2026-9317: upgrade nango to 0.71.6 and set NANGO_INTERNAL_AUTH_REQUIRED=true
HOL Guard
CVE-2026-85024: undici WebSocket deflate bug can crash the Node process
How to fix CVE-2026-85024: upgrade undici to 8.10.2 (or 7.29.1 / 6.28.1 on older trains)
HOL Guard
BREAKING: CVE-2026-76169 lets malformed URLs skip Fastify not-found auth
How to fix CVE-2026-76169: upgrade fastify to 5.12.2
HOL Guard
BREAKING: CVE-2026-85184 lets absolute-form requests skip Fastify middie auth
How to fix CVE-2026-85184: upgrade @fastify/middie to 9.3.4
HOL Guard
BREAKING: CVE-2026-85046 is a Chrome V8 bug Google says is exploited in the wild
How to fix CVE-2026-85046: upgrade Chrome to 152.0.7977.82 (Linux) or 152.0.7977.82/.83 (Windows and Mac)
HOL Guard
CVE-2026-71963: Hermes Agent runs Git config before the first prompt
How to fix CVE-2026-71963: update Hermes Agent to a build containing commit f6234d0 or a later vendor release.
HOL Guard
CVE-2026-75033: one Rancher annotation copies another cluster's secrets
How to fix CVE-2026-75033: upgrade Rancher to 2.15.1, 2.14.5, 2.13.9, or 2.12.13, and move rancher-webhook with it.
HOL GuardHOL Guard research desk
Security research for the AI agent era
Threat guides and evidence dossiers on prompt injection, MCP tool poisoning, slopsquatting, and the attacks shaping how teams ship code with agents.
Prompt injection
How hidden instructions trick coding agents into unsafe work.
Read the researchEvergreen guideMCP security
Tool poisoning, overbroad permissions, and shadow MCP servers.
Read the researchNew dossierSlopsquatting and package hallucination
Slopsquatting is the supply-chain attack in which attackers register package names that AI models reliably hallucinate; when an agent writes an import for a package that never existed, the attacker’s registered code installs instead.
Read the researchNew dossierMalicious extensions and marketplace fraud
Malicious extensions and marketplace fraud place trojanized editor extensions, plugins, or CI actions inside trusted marketplaces, harvesting source code, prompts, and credentials from every developer who installs them.
Read the research