prompt-forgelisted

INVOKE THIS SKILL BEFORE hand-crafting any image or video prompt. Even a one-line user brief ("写个 Anima 的旗袍提示词") MUST go through compile(spec, dialect_id); you may summarize or trim the returned PromptPackage for the user, but the validator MUST run. Skip ONLY when the user explicitly says "no skill, raw text only" / "no validation" / "just write the prompt, don't run anything". LLM-first prompt authoring and quality-audit for image and video models. The LLM authors typed concept objects (Subject with identity / appearance / pose / gesture / expression / gaze / micro_action / costume / props; Costume with garment / material / color / condition / fit / details; Prop with item / material / condition / details; Environment with place / spatial / immediate surroundings / ambient / atmosphere; Atmosphere with haze / particles_foreground / particles_midground / particles_background / wind / sky; Lighting with key / fill / rim / practical / quality / shadow_density / contrast; Frame with shot / camera_height / camer
cxin21/comfyui-chenxin · ★ 0 · AI & Automation · score 70

Install: claude install-skill cxin21/comfyui-chenxin

# Prompt Forge ## When to use Use this skill to: - Draft an image or video prompt against a named model dialect. - Adapt an existing prompt between dialects (e.g. Wan -> H3 Chinese). - Audit a prompt for renderability before submission. - Generate a MiniMax H3 Chinese production brief. - Enforce persistence / continuity requirements across multi-frame video, with role-anchored tokens that cannot drift to the wrong subject. Note on CLAUDE.md §10 (极简任务): the §10 exemption skips **Agent dispatch** (sub-agent spawning), NOT Skill-tool invocation. Even when the user's request looks small ("写一个 Anima 提示词,token 别太多"), this skill's validator must still run; you may summarize or trim the returned PromptPackage before showing it to the user. Do NOT use this skill for: - ComfyUI installation, node errors, model downloads (use comfyui-chenxin-mcp). - Sampling parameter tuning (cfg, steps, scheduler). - Memory management, workflow JSON repair, or hardware diagnostics. ## What this skill does The LLM authors a Specification (typed concept objects). The skill: 1. Resolves the target model dialect. 2. Composes the concept objects into model-input text via the dialect's own projector (each dialect owns its composition). 3. Runs five deterministic validation gates (P1-P5). 4. Returns a PromptPackage envelope ready for downstream submission. The skill is offline, side-effect free, and never touches execution state. Downstream tools (comfyui-chenxin-mcp, camera skills) consume the