An evidence-led, eight-locale LLM playbook: a transferable core, the Codex flagship track, and adapters for ChatGPT, Claude Code, Gemini, DeepSeek, and Grok.
This repository also configures its own agents. See what Prysai-LLM-Playbook tells them →
Skill Codex
Audit, draft, and revise public-facing prose for open-source software so it is specific, credible, and natural without flattening the project's voice. Use for README files, documentation, landing pages, release notes, changelogs, contribution guides, PR or issue text, UI copy, error messages, prompts, Traditional…
Skill Codex
Review an LLM learning product, documentation site, Skill library, or candidate release from the strongest plausible opposing case. Use when deciding whether a project is genuinely useful, safe, teachable, maintainable, or ready to publish; when a team asks for a professor, scientist, practitioner, or open-source…
Skill Codex
Coach a learner through GPT, Codex, tools, skills, Agent workflows, verification, and team practice. Use when someone asks how to learn Codex, improve a Codex workflow, understand observable Agent behavior, practice a capability, or judge whether they have mastered a task. Do not use as the primary router for an…
Skill Codex
Diagnose an already failed LLM interaction from the original request, visible context, actual reply or artifact, and expected outcome; propose the smallest communication repair and a controlled rerun. Use when a reply ignored constraints, answered the previous task, caused repeated rework, or remained impossible to…
Skill Codex
Turn an untried, low-risk request for LLM help into one short, copy-ready first-turn dialogue brief with a stated outcome, audience, supplied inputs, constraints, output shape, acceptance check, and stop boundary. Use when the user has a roughly defined communication goal but needs a clear first message; do not use…
Skill Codex
Audit Codex, Agent, research, marketing, browser, deployment, skill, or task-completion claims against observable evidence. Use when a result may be polished but incomplete, when status must be separated into verified, inferred, blocked, or unknown, or when a smallest next check is needed. Do not use to execute the…
Skill Codex
Convert public forum posts, issues, support discussions, interviews, or repeated user questions into a traceable demand and failure-signal record. Use when deciding what an LLM tutorial, platform adapter, Lab, FAQ, or application playbook should teach. Do not use community reports to prove a root cause, official…
Skill Codex
Inspect a user-written, unsent, text-only, low-risk first request for six visible fields: outcome, starting context, requested response, limits, check, and stop-and-receipt. Use when the user wants a compact before-send review rather than a new prompt, coaching session, research task, file/tool plan, or failed-reply…
Skill Codex
Preserve observable task state after an interrupted LLM-assisted task and choose one safe next decision. Use when a model is unavailable, a task times out, a session is lost, a tool is missing, or a handoff disconnects before acceptance evidence is visible. Do not use to retry, diagnose a preserved interaction, audit…
Skill Codex
Rehearse one observable interview answer under a time limit: the candidate answers first, the coach names one material gap with a partial cue, the candidate revises, then answers one changed question unassisted. Use when a person says "help me prepare for an interview", "rehearse an answer about my project", or "I…
Skill Codex
Run one bounded typed exchange in the learner's target language: the learner writes first, the partner plays one native-speaker role, corrects at most one meaning-blocking error with a partial cue, and later runs one changed case. Use when a learner says "practise Spanish with me", "rehearse a French study group…
Skill Codex
Turn a non-Codex skill the user wants to learn into a short practice loop built around baseline assessment, attempted recall, corrective feedback, spaced review, and transfer. Use for language learning, software concepts, writing, interviewing, or other learnable skills when the user wants an AI tutor, practice…
Skill Codex
Plan or review a fair, task-scoped comparison between two named LLM, model-provider, or workflow candidates. Use when someone asks which option is better, faster, cheaper, or worth expanding and the answer must preserve fixed conditions, raw evidence, unavailable cases, and a narrow conclusion. Do not use to admit a…
Skill Codex
Audit a proposed tutorial or workflow for Codex, Claude Code, Grok, ChatGPT, Gemini, Copilot, or another LLM platform against a platform-adapter contract. Use when deciding whether platform-specific content adds a sourced, runnable, maintainable delta from the universal LLM core. Do not use for a general model ranking…
Skill Codex
Maintain current, named-platform teaching claims after a vendor document, product surface, permission, model, account path, or link may have changed. Use when a curriculum maintainer needs to identify affected chapters, Labs, Skills, routes, and temporary reader-facing limits for Codex, Claude Code, Grok, ChatGPT…
Skill Codex
Record one low-risk, user-authorized first-use observation of a named LLM platform or surface without inferring capability, parity, safety, or success. Use when a learner has opened Claude Code, Grok, ChatGPT, Gemini, Copilot, Codex, or another platform and needs an evidence receipt for what was visibly offered…
Skill Codex
Turn an ambitious or vague learning wish into one small, honest, prompt-ready practice target. Use when a learner says "learn Spanish in seven days", "get better at interviews", "learn a skill with AI", or asks where to start a time-limited learning goal. Set one situation, baseline, session budget, allowed help…
Skill Codex
Create or update a versioned product and marketing context before positioning, content, SEO, conversion, launch, analytics, or sales work. Use when shared product understanding is missing or audience, positioning, brand voice, or product context is requested. Do not use to invent customer evidence, replace research…
Skill Codex
Convert a project-authored or explicitly authorized prompt idea into one beginner-facing, copy-ready teaching card with a stated task, supplied context, action limits, self-check, recovery route, and source boundary. Use when maintaining a prompt-card library, turning a reviewed lesson idea into a reusable card, or…
Skill Codex
Route one incoming LLM request to the smallest safe next method before drafting, researching, or acting. Use when a beginner is unsure whether a request is a supplied-text draft, one current fact, multi-source research, or an external action or change. Return a routing receipt only; do not execute, look up sources…
Skill Codex
Route research, literature review, fact checking, comparison, academic writing, and source-backed reports through question scoping, source planning, retrieval, evidence extraction, synthesis, citation, disclosure, and review. Use when a request needs sources or has an unresolved research question. Do not use for…
Skill Codex
Prepare one current-work brief for a recurring LLM collaboration by keeping reusable criteria, the changing item, authority, and acceptance evidence distinct. Use when yesterday's context or an earlier example could be mistaken for today's work. Do not use to create a product context, design a full task protocol…
Skill Codex
Select, compare, install, or combine the smallest useful Codex skill set for a specific task. Use when someone asks which skill to use, considers a skill catalog, or needs to assess trigger fit, dependencies, permissions, license, maintenance, and rollback. Do not use for general learning, evidence-only review, source…
Skill Codex
Execute a narrow current-source investigation with a precise question, source hierarchy, claim ledger, conflict handling, freshness checks, and a stopping rule. Use when the user asks to find, verify current information for a decision. Use Research Router instead to scope a broad topic, design a literature review, or…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: