OpenLink · AI Agent Skills · Screencast

Establishing Session Context from an RDF Memory System

A narrated walkthrough of a real Grok CLI session: how the agent loads agent-rdf-memory, applies standing preferences, verifies WebID identity, and scopes prior history before continuing work.

Principal Kingsley Uyi Idehen Agent Grok 4.5 · Grok CLI On behalf of principal (agent-only mode) Date 2026-07-23 Duration 2:20 Skill screencast-recorder

Primary media: H.264 MP4 with AAC voice-over (coral TTS). Poster frame from the title scene. Silent WebM/MP4 retained as alternate encodings.

What this shows

Context is established as a gated procedure, not inferred from chat alone. The session replays three user turns against the live memory contract on the ai-agent-skills repository.

Memory protocol

List agent-rdf-memory/, then read core.ttl, preferences.ttl, and index.ttl before answering.

Verified whoami

Trigger A runs modulus checks and reciprocal delegation corroboration; badges are earned.

Scoped history

Session reads default to this model (Grok). Other agents stay out until the filter is lifted.

Session beats

  1. session context? — Execute the Agent RDF Memory Protocol; report runtime, branch, bootstrap gates, artifact root.
  2. whoami? — Verified identity gate: credentials root, user/agent modulus MATCH, local + remote reciprocal delegation.
  3. Prior session history? — Grok-only filter (user choice): thin Grok 4.5 trail, denser Grok 4.3 cluster; 124 multi-agent files not loaded.

Production facts

Duration
2:20
Resolution
1440×900
Voice
coral TTS
Contract steps
154

Recording pipeline

HTML scene deck (presentation-scenes-recording-deck.html) → shot-scraper video silent record → OpenAI TTS MP3 → ffmpeg mux to final MP4. Bundle lives under the Grok screencasts/ output root (not webpages/).

Narration

Full voice-over script used for the coral TTS track.

Show narration.txt 
Loading…