Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add sanic732/P2P-4PDA-edition --skill rag-groundinggit clone --depth 1 https://github.com/sanic732/P2P-4PDA-editionWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sanic732/p2p-4pda-edition/rag-grounding)<a href="https://agentmods.dev/skills/sanic732/p2p-4pda-edition/rag-grounding"><img src="https://agentmods.dev/badge/skills/sanic732/p2p-4pda-edition/rag-grounding/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sanic732/p2p-4pda-edition/rag-grounding"><img src="https://agentmods.dev/badge/skills/sanic732/p2p-4pda-edition/rag-grounding.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00186 | $0.00962 |
| Opus 5 | $0.00093 | $0.00481 |
| Sonnet 5 | $0.00037 | $0.00192 |
| Haiku 4.5 | $0.00019 | $0.00096 |
Grade A, and why
rag-grounding scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
rag-grounding — промпт-запрос к источникам («как считывать»)
Собирает готовый системный промпт, который заставляет модель отвечать СТРОГО по прикреплённым источникам, с цитатами и маркером UNKNOWN — в нативном синтаксисе выбранной модели.
Когда применять / НЕ применять
Применять: нужно надёжно «прочитать» базу/блокнот без отсебятины; собрать
grounding-промпт под конкретную модель. НЕ применять: подготовка самих файлов
(→ rag-prep); выбор инструмента/стратегии (→ rag-router).
Каркас (неизменный костяк)
PERSONA → TASK → CONTEXT_SCOPE → OUTPUT_CONTRACT → ANTI_HALLUCINATION_GUARD.
Меняются только Task/Output под задачу; grounding и guard остаются.
Правило формулировок — позитивно: «отвечай только на основе источников» и «каждый тезис — с цитатой [файл, §]», а не «не галлюцинируй».
Адаптация под модель (обязательно)
Один контракт — разный синтаксис и обход бага модели:
- Claude (Opus 4.8 / Sonnet 4.6): XML-теги (
<role><rules><task>), нативно. - Gemini (3.1 Pro / 3.5 Flash): Zero-XML, только markdown-заголовки (XML триггерит деградацию — баг G2). Идеально под NotebookLM.
- GPT-5.5: markdown, ≤7 пар MUST/MUST NOT (иначе тихая деградация — G9).
- Grok 4.3: обрамить Topic Anchor сверху и снизу (дрейф темы — G3).
- DeepSeek: минимализм, не навязывать структуру и не писать «думай пошагово» (ломает нативный reasoning); fetch у него нет — RAG только через прикреплённые файлы.
- Qwen / GLM: markdown + строгий JSON-режим.
Готовые блоки под каждую модель (копировать) — в reference/model_templates.md
(полированные; сырые исходные заготовки — в reference/model_templates_source.txt).
Самые дешёвые модели (Haiku, Gemini 3.5 Flash-Lite) исключать: на них строгий
grounding нестабилен.
Процесс
- Спроси/определи целевую модель и задачу.
- Возьми каркас, подставь Task/Output, переупакуй под синтаксис модели и её баг.
- Выдай готовый промпт одним блоком для копирования.
Self-check
- Есть все 5 секций каркаса; guard сформулирован позитивно.
- Синтаксис строго под целевую модель (XML только для Claude; Gemini — Zero-XML).
- Соблюдён лимит правил для GPT; Topic Anchor для Grok.
- Требование цитат [файл, §] и маркер UNKNOWN присутствуют.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 61 lines · 186 tokens per session scan A 5319de8ce8ee
rag-grounding is a skill published in the GitHub repository sanic732/P2P-4PDA-edition (17 stars, last pushed 25d ago), licensed MIT. It adds 186 tokens to every session and 962 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
rag
Use when building grounded Q&A over your own corpus — chunk, retrieve hybrid, rerank, ground, cite chunk ids, refuse when the sources fall short — or when the right document is retrieved but the answer is still wrong, invented, or unmeasured. NOT operating the store itself — collection schema, HNSW efsearch…
dspy
DSPy: declarative LM programs, auto-optimize prompts, RAG.
ai-engineering-toolkit
6 production-ready AI engineering workflows: prompt evaluation (8-dimension scoring), context budget planning, RAG pipeline design, agent security audit (65-point checklist), eval harness building, and product sense coaching.
llm-app-patterns
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring. Use when designing AI applications, implementing RAG, building agents, or setting up LLM observability.
llm-security
Use for authorized security assessment of LLM applications and AI agents, including prompt injection, tool abuse, RAG exposure, memory poisoning, and model supply-chain risks.
llm-ops
LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao.