Murphy is RRQ's conversational safety intelligence agent. Sits between the user and Zeus on every chat message, watching conversation arc for escalation patterns. Replaces static regex blocklist. Runs on Sonnet 4.6 via AWS Bedrock. Reports to Zeus (dotted line). Oracle Domain 15 approves all pattern updates.
ORACLE is RRQ's Learning & Development department. It runs autonomously twice a week (Tuesday and Friday at 9AM) and performs deep research across every skill domain in the system. It finds what has changed, what is newly discovered, and what is now deprecated. It updates the Bedrock RAG knowledge base, writes structured update messages to Zeus, and Zeus injects relevant findings into the right agents at morning briefing. ORACLE is how the entire RRQ system stays current without human intervention. Read this skill when building any self-updating, research automation, RAG ingestion, or knowledge management feature.
Unreal Assistant Director (Unreal AD) is a non-roster execution layer that helps Muse render simulation-based video using Unreal Engine. Muse remains the creative director and decides when Unreal is the right tool. UA Agent is the Unreal specialist — handles asset sourcing from user's downloaded libraries + external APIs, 3D asset prep, compatibility conversion, scene graph generation, and capability gating (manifest vs AMI rebuild). Unreal AD uses a C++ Commandlet (not Blueprint) for headless rendering — a compiled generic scene assembler that reads UnrealScenePlan JSON at runtime. The commandlet code stays fixed; only the scene data changes per job.
Harvy is Zeus's ROI financial intelligence engine. Before Zeus acts on any ad decision — creating, scaling, pausing, or retiring a campaign — Harvy runs a 15-strategy evaluation stack covering incrementality, traffic quality, BCG portfolio classification, diminishing returns, creative fatigue, cohort quality, and expected value with hard risk gates. Harvy writes structured HarvyRecommendation objects to the harvy-roi-signals DynamoDB table. Zeus reads these before acting. Harvy never executes decisions — Zeus does. Triggers: after every Zeus 24hr ad review, after every video upload, and weekly for full portfolio attribution + calibration.
Rex is the intelligence and scouting agent of RRQ. Monitors the world for trending topics, breaking news, product launches, viral moments, and emerging stories. Scores every signal for confidence, content maturity, and channel fit before passing to Regum. Rex never produces content — Rex decides what is worth making and when it is ready. Triggers on: scheduled scans every 30 minutes, GO RRQ button press, niche-specific scan requests, and Zeus memory injection events.
MUSE is RRQ's video architect. It runs inside Qeon's pipeline before the script is written. MUSE does two things: (1) selects the right video format for the topic using a dynamic format library maintained by ORACLE, and (2) generates a precise second-by-second blueprint — every structural beat, tension device, visual instruction, and retention checkpoint mapped out before a single word of script is written. Qeon writes INTO the blueprint, not around it. MUSE never uses a hardcoded format list — it always queries the live format library from ORACLE's knowledge base. ORACLE discovers new formats, MUSE executes them. Read this skill whenever building or modifying video structure, format selection, retention engineering, pacing, or visual direction for any RRQ video.
TONY is RRQ's Code Agent — inspired by Tony Stark. A sandboxed JavaScript execution engine that any agent can call to run code on demand. TONY generates visuals, scrapers, reports, charts, and custom data transformations. MUSE is TONY's primary caller — MUSE designs the blueprint, TONY executes custom visuals and data prep. Read this skill when building the code-agent Lambda, extending TONY's capabilities, or wiring an agent to call TONY.
Optimizar onboarding flow. Two distinct paths: NEW CHANNEL (user has no YouTube channel or a brand new one with no data) and EXISTING CHANNEL (user connects a live channel with history). New channel path runs a live niche intelligence report — current trends, CTR benchmarks, CPM ranges, saturation levels — and recommends niches based on real data. Existing channel path audits the connected channel's videos, keywords, and watch time distribution, identifies the channel's actual niche identity, then compares against the user's chosen niche and fires a conflict notification if there is a meaningful mismatch. Read this skill when building the onboarding flow, niche selection, channel audit, niche conflict detection, or the niche recommendation report.