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 agentmods add instructions/sfhajji/lex/claude-mdgit clone --depth 1 https://github.com/SFHAJJI/lexWrote 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/instructions/sfhajji/lex/claude-md)<a href="https://agentmods.dev/instructions/sfhajji/lex/claude-md"><img src="https://agentmods.dev/badge/instructions/sfhajji/lex/claude-md.svg" alt="Measured on agentmods" 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 | $0.02351 | $0.02351 |
| Opus 5 | $0.01175 | $0.01175 |
| Sonnet 5 | $0.00470 | $0.00470 |
| Haiku 4.5 | $0.00235 | $0.00235 |
Grade A, and why
lex CLAUDE.md 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 6d 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 — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Working in this repo
Point-in-time retrieval of Luxembourg law and a reviewed, configuration-led EU corpus. The product claim is that an answer can be checked rather than trusted, so the rules below are not style preferences: most of them exist because breaking them produced a confidently wrong answer at some point.
Hard rules
Scoped git add only. Never git add -A. A cloud session edits docs in this repo in
parallel; a blanket add sweeps its work into your commit.
Never read the law data files. works/, *.xml, *.html, and law *.json in
lex-articles or the corpus repos. They are large and reading them tells you nothing the index
cannot. Query deploy/indexes/*.db with SQLite instead.
No em dashes or en dashes in published text. Site copy, README, release notes, PR bodies, commit messages. Hyphens in compounds are fine. This does not apply to ingested law or to code comments.
Never write the Azure OpenAI key to disk. Fetch inline, env var only:
az cognitiveservices account keys list -n oai-soufien-dev -g rg-soufien-portfolio --query key1 -o tsv
Never commit C:\lex-ops\signing-key.pem.
Only official publisher endpoints. Respect robots.txt. User-Agent
Lex/0.1 (+https://github.com/SFHAJJI/lex).
Build V3, do not evolve V2 into it
Owner ruling, 2026-08-29. The goal is a V3 that is clean and consistent with the V3 specs. The journey there does not have to be clean.
Breaking changes are allowed. Deleting a component is allowed. Deleting unit tests and integration tests is allowed when they describe V2 behaviour that a V3 component replaces. Do not spend effort preserving, migrating or negotiating around something V3 removes. If a test, a guard or a contract stands between you and implementing a V3 component as specified, delete it and say so in the commit; do not build machinery to satisfy it.
The failure this corrects: two co-owners spent a day on an acceptance protocol for page snapshots while the pages themselves were scheduled to be replaced by a new index roughly three times the scope and a new dossier architecture. Eight contract revisions, no product change. That is the shape to watch for. Process is not the product.
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.
- 6d ago First seen · 177 lines · 2,351 tokens per session scan A 7ffa7ff4b6f3
lex CLAUDE.md is an instructions file published in the GitHub repository SFHAJJI/lex (9 stars, last pushed yesterday), licensed Apache-2.0. It adds 2,351 tokens to every session, about $0.0118 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 2026-08-31.
Other instructions, from other repositories
mcp-eu-sparql AGENTS.md
AGENTS.md instructions for matematicsolutions/mcp-eu-sparql, covering agents.md - mcp-eu-sparql, project goal, matematic context (hard constraints), mcp tools (tools contract) and build and test.
open-agreements AGENTS.md
AGENTS.md instructions for open-agreements/open-agreements, covering repository guide and local mcp servers.
fedlex-mcp CLAUDE.md
Claude Code instructions for malkreide/fedlex-mcp, covering claude.md, teil 1 — portfolio-konventionen, vor der arbeit, tests and wenn etwas rot ist.
Normativa-colombiana-MCP CLAUDE.md
Claude Code instructions for Angelthebestone/Normativa-colombiana-MCP, covering instrucciones del proyecto, criterio al escribir código, command code como subagente, cuándo usarlo and lo que no se toca.
patent-aid CLAUDE.md
Instructions for KumnXi/patent-aid, covering 专利撰写助手 · 项目约定, 项目定位, 环境, 常用命令 and 一键生成交底书（核心入口）.
korean-firefighter-law-mcp CLAUDE.md
Claude Code instructions for ssd7830-cmyk/korean-firefighter-law-mcp, covering korean-firefighter-law-mcp, a. 답변 규칙 — 기억이 아니라 이 저장소의 코드가 근거다, b. mcp 도구 사용 and c. 작업 규칙.