Dallionking

Take a rough goal description and shape it into a polished, doctrine-compliant agent goal prompt. Output is for the user to copy/paste into Codex, Claude Code (native /goal), Droid, Auggie, or any CLI agent that benefits from a structured goal contract. The skill does NOT execute the goal — it shapes the prompt and stops. Triggers on "/agent-goal-skill", "/goal-shape", "shape this goal", "make this a goal prompt", "turn this into a /goal block".

Prepare and execute your hardest, highest-leverage work for your most capable model. ONE adaptive loop. When the frontier model (Fable 5) is UNAVAILABLE it runs PLAN-MODE — sweeps your repos across every domain (code, design, copy, marketing), adversarially verifies each finding, and emits a globally-ranked WAVE QUEUE of execution-ready packets. When the frontier model is LIVE it runs EXECUTE-MODE — straight-executes the queue through tiered gates, pausing per wave. The thesis: the frontier model one-shots what cheaper models grind through rounds — so you never spend its scarce, expensive time on planning; you spend it shipping. Triggers on 'fable prep', 'prep for fable', 'fable queue', 'fable night', 'prep hardest tasks', 'round-collapse', 'prep for my best model', 'frontier queue'.

Lion's Paw Academy student context — curriculum, tier gating, module map, Sigma Protocol vocabulary, tier-gated GitHub repos, Discord community, and support paths. Use when working through any Lion's Paw Academy lesson, following a lesson prompt, or building a project alongside the curriculum. Triggers on 'lion's paw', 'lionspaw', 'academy lesson', 'M0'-'M7', 'M1.5', 'T1'-'T7', 'LP0'-'LP8', 'sigma protocol', 'harness engineering', 'agentic engineering', 'quantstream', 'paw credits', 'the den'.

Walk the user through running GPT, Grok, and other models INSIDE the Claude Code harness via CLIProxyAPI — install, OAuth account connect (ChatGPT/Codex, X Premium/Grok), launcher scripts, verification, and optional Orca multi-agent dispatch. Use when the user says "set up cli proxy", "run gpt in claude code", "use grok in claude", "multi-model claude", or "one harness every model".