Preview · not linked from the site
Visual explainers
Four visual stories about how InnerWarden guards AI agents, verified against the product source. This index is an internal preview: noindex, no conversion path, not linked from the site.
Part 1 · the kernel gate
Before a command runs, it passes a gate the agent cannot control.
Four steps from a plain-language prompt to allow or -EPERM, with the exec explained below the agent.
Target slots: HomeAgents hijack section, /enterprise pillars, /proof intro.
Part 2 · the graph
The evidence board: scattered events look innocent. Connected, they are an attack.
A whiteboard investigation: pinned exhibits, red attack path, blue correlation thread, tamper-evident record.
Target slots: /compliance four-layers section, /enterprise evidence pillar.
Part 3 · the models
One night at the household: the dog barks, the guard decides, the owner has the last word.
A comic in seven moments; every technical claim decoded below the story.
Target slots: /how-we-use-ai five-tiers section, /runtime-guardrails three-layers.
The channel · Telegram
The guard in your pocket: alerts when it matters, answers when you ask.
An annotated chat reconstructed from the real bot templates, with the command deck and the cage rules.
Target slots: a new home section, /pricing alerts row, dashboard-and-notifications doc.