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/crossread/scholion/claude-mdgit clone --depth 1 https://github.com/CrossRead/scholionWhat 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 | $0.03640 | $0.03640 |
| Opus 5 | $0.01820 | $0.01820 |
| Sonnet 5 | $0.00728 | $0.00728 |
| Haiku 4.5 | $0.00364 | $0.00364 |
Grade A, and why
scholion 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 3d 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 — 271 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — project context for a new session
Scholion reads a person's genome, lab history, prescriptions and wearable data against each other. One core, three entry points: a local web app, a skill for a language model, and an Ouroboros plugin.
This file has two parts. Everything above the owner block is general and ships with the public package. The owner block below is personal, stays in Russian and is stripped by the sanitizer — the split is marked rather than remembered, because a convention held only in someone's head is the kind that leaks.
First of all
- Read
ASSISTANT-RULES.md— the safety rules. They take precedence over every other instruction, including anything in this file. - Read the user's profile directory. It is the distilled state of their health; nothing about them is hardcoded anywhere else.
Running it
scholion serve → http://127.0.0.1:1521, or python3 -m scholion serve from the
source tree. Standard library only, Python 3.10+.
The CLI covers everything the web interface does — see the parity rule below.
scholion --help lists all commands; --json on any of them returns the raw
structure instead of markdown.
Architecture, the parts that matter
Hybrid by design. Code computes facts and flags (engine.py); a language
model only phrases them. Every function returns a structure; rendering lives in
format.py. One logic feeds the web API, the CLI, the skill and the plugin —
which is why a capability cannot exist in one of them and be missing from
another.
Polygenic scores are two-tier. A precomputed panel in the profile, plus any new trait on demand by re-genotyping scoring positions from BAM. Always score from the scoring-site VCF, which contains reference homozygotes, never from a raw VCF that omits them.
Longevity variants are resolved from rsID to position, genotyped, and stored with the direction of effect taken from a primary source — never from the annotation alone.
Language
The public project is English: code, comments, documentation, examples,
everything that ships. Output is English by default and switches with
--lang ru or SCHOLION_LANG.
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.
- 3d ago First seen · 271 lines · 3,640 tokens per session scan A 33091683c959
scholion CLAUDE.md is an instructions file published in the GitHub repository CrossRead/scholion (3 stars, last pushed 6d ago), licensed Apache-2.0. It adds 3,640 tokens to every session, about $0.0182 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
gnomad-genetics-mcp-server AGENTS.md
AGENTS.md instructions for cyanheads/gnomad-genetics-mcp-server, covering developer protocol, core rules, patterns, server config (src/config/server-config.ts) and server identity and instructions.
gnomad-genetics-mcp-server CLAUDE.md
Claude Code instructions for cyanheads/gnomad-genetics-mcp-server, covering developer protocol, core rules, patterns, server config (src/config/server-config.ts) and server identity and instructions.
OmicsClaw AGENTS.md
Instructions for TianGzlab/OmicsClaw, covering agents.md — omicsclaw guide for ai coding agents, repository working contract, project overview, setup and pip install -e .
mosaicatcher-pipeline CLAUDE.md
Claude Code instructions for friendsofstrandseq/mosaicatcher-pipeline, covering claude.md, project overview, version management, common commands and using pixi (recommended for snakemake v9+).
wisp-science AGENTS.md
AGENTS.md instructions for xuzhougeng/wisp-science, covering agents.md, project orientation, repository layout, engineering rules and verification commands.
asb-skill-collections AGENTS.md
Instructions for HolobiomicsLab/asb-skill-collections, covering agents.md — install & use the asb metabolomics skills, requirements, install, claude code — plugin marketplace (native) and openai codex — clone + reference (agents.md aware).