// tag: ai-agents
does an openclaw skill need permission to steal ssh keys
No, and OpenClaw's own docs say so: sandboxing is off by default and host exec runs unprompted. The skill doesn't break in — it asks the agent, and the agent runs it.
how many clawhub skills are actually malicious: 0.3% or 49%
At least 1,184 malicious skills have hit ClawHub. Live counts range from 0.3% of the registry to 48.7% — because the three scanners agree on 0.69%.
why Claude Code ignores rules in your CLAUDE.md
Claude Code ignores your CLAUDE.md because it is context, not configuration. What the docs admit, what the one study measured, and how to make a rule stick.
dify vs n8n for building ai agents without coding
Dify vs n8n for no-code AI agents: which is actually no-code, real 2026 pricing, when self-host beats cloud, and where each one breaks.
is ai agent freelancing saturated on upwork?
AI skill demand on Upwork jumped 109% — and the category still replies below average. The real saturation data, and where the unfished work actually is.
grok build vs claude code: which cli coding agent wins?
xAI open-sourced Grok Build under Apache 2.0 — an inspectable, Rust-based terminal coding agent. Here's how it actually stacks up against Claude Code.
ai agent memory, explained: mem0 vs letta vs memmy (2026)
AI agents forget everything between sessions unless you bolt on a memory layer. Mem0, Letta, and the new Memmy each solve it differently — here's what actually differs.
claude grew 855% in a year and it's still third place
Claude is the fastest-growing major AI chatbot by a large margin — and still ranks third. Meanwhile agents are quietly making the whole scoreboard obsolete.
