532 posts tagged with "Security"
Cybersecurity, smart contract audits, and best practices
Harbor's CVSS 9.4 Default Password: How to Tell If Your Registry Is Exposed, and the First-Boot Rule That Prevents the Next One
Harbor 2.15.0 and below ships with a working admin/Harbor12345 default worth CVSS 9.4. A five-minute exposure audit, the rotate-first fix order, what to hunt if it was reachable, and the first-boot rule your own platform should steal.
SEP-835 Scopes: Least-Privilege MCP Tokens for Agents That Hold Deploy Keys
MCP's SEP-835 adds native per-tool scopes and RFC 8707 audience binding, so a status-check agent holds read-only scopes while deploy and rollback stay behind explicit step-up — and a compromised tool call fails closed instead of shipping an unauthorized deploy.
Render's CLI Now Phones Home by Default: What Opt-Out Telemetry Means for Trusting Your Deploy Tool
Render CLI v2.26.0 collects usage telemetry by default. This breakdown covers exactly what it sends, how to opt out, the CI disclosure gap in its notice design, and the standard credential-holding CLIs should meet.
Stop Passing Service Account Tokens to CSI Drivers in Volume Context: What the v1.35 Token-in-Secrets Fix Means for Your Storage Layer
CSI drivers received workload-identity tokens in a log-visible gRPC map, and two CVEs printed them into driver logs. How the v1.35 secrets-field fix works, the rollout order that avoids breaking mounts, and how to audit your fleet.
Synacktiv's Unpatched Argo CD Repo-Server Flaw: Why Any Pod That Can Reach the gRPC Service Is Equivalent to an Authenticated Attacker
Argo CD's repo-server exposes an unauthenticated gRPC endpoint that turns any pod with network reachability into a path to cluster takeover. This breakdown covers Synacktiv's exploit chain, which installs are exposed, and the exact NetworkPolicies that close it.
A2A Joins MCP Under One Roof: How to Hand an Ops Agent a Deploy Investigation Without Handing Over Deploy Authority
A2A joining MCP under the Agentic AI Foundation settles which protocol does what: A2A moves the diagnosis between agents, MCP moves capability between an agent and the platform. Here is the handoff design that keeps it that way — a worked deploy-investigation task, a three-tool audited MCP surface, and the identity, approval, and least-privilege invariants mapped to Kubernetes RBAC.
Your Agent Can Call 7,000 SaaS APIs. Can It Deploy to a Machine You Own?
Arcade.dev gives agents 7,500+ SaaS tools behind brokered OAuth; Capix MCP gives them 37 tools that deploy models, provision GPUs, and emit signed receipts. Only one of those surfaces operates infrastructure you own — here is the checklist your PaaS MCP server still has to complete.
A Czech Self-Hosting Hub Gave Claude Shell Access to Its Coolify Fleet: A 130-Line MCP Bridge Teardown
A Czech self-hosting hub wired Claude to its Coolify fleet through a 133-line MCP bridge — and handed the agent raw SSH instead of scoped deploy tokens. A code-level teardown of all seven tools, plus the five authorization rules a platform-owned MCP server needs.
Chat Deploys My App: What the Coolify–AnythingLLM MCP Bridge Proves, and the 5 Gaps a Render-Compatible MCP Server Closes
A community MCP server lets an LLM inspect, restart, and debug apps on a self-hosted Coolify box through AnythingLLM chat — but behind one shared API token sit a docker-socket tap and a root shell over SSH. Here is what the bridge proves, how Render's official MCP server scopes agent power instead, and the five gaps a Render-compatible MCP server still has to close.