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/shigella520/mindtrain/agents-mdgit clone --depth 1 https://github.com/shigella520/MindTrainWhat 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 | $0.00559 | $0.00559 |
| Opus 5 | $0.00280 | $0.00280 |
| Sonnet 5 | $0.00112 | $0.00112 |
| Haiku 4.5 | $0.00056 | $0.00056 |
Grade A, and why
MindTrain AGENTS.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 2d 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 — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Instructions
MindTrain target architecture
- Treat MindTrain as a domain-independent knowledge-training platform, not as a Java-only Skill.
- Use
doc/目标需求.mdas the product requirements source anddoc/概要设计.mdas the architecture source. - Keep Training Core as the source of truth for questions, candidates, prompts, sessions, attempts, review events, scheduler bindings, and plugin sync state.
- Keep Codex Skills stateless: they orchestrate MCP/API tools and must not become the authoritative application database.
- Keep scheduler behavior behind a provider contract. The initial providers are Core 加权调度 (stable ID:
weighted) and the optional Anki scheduler plugin. - Treat Anki as a scheduling plugin and rebuildable local projection, not as the authoritative MindTrain question bank.
- Keep production runtime data out of the repository; retain only migrations, contracts, deployment configuration, and minimal non-private test fixtures.
- Do not introduce microservices prematurely. Build Training Core as a modular monolith first.
Training and content integrity
- Use
plugins/mindtrain/skills/mindtrain/SKILL.mdas the only MindTrain Skill source. - Access questions, candidates, sessions, attempts, mastery and reports only through Trainer MCP or Training Core APIs.
- Permit an unanswered generated candidate only in its owning session; answering activates it, while explicit rejection physically deletes it.
- Grade single- and multiple-choice answers by exact option-set equality; scores are only 100 or 0.
- Never show an example answer combination that could disclose the correct answer count or option set. Use neutral text such as
请回复选项字母，可用逗号分隔. - Record source URL, access date, version, and review state for generated questions.
- Increment the version when published question content or scoring criteria change.
- Use test-local fixtures rather than repository question banks or personal learning files.
Safety and integrity
- Do not commit secrets, API keys, Anki credentials, access tokens, or private learning material.
- Do not overwrite or delete existing user learning data while restructuring the platform.
- Make external integrations idempotent and retain enough state to recover partial Core/plugin failures.
- Keep third-party MCP and Anki Add-on APIs behind MindTrain-owned adapters.
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.
- 2d ago First seen · 38 lines · 559 tokens per session scan A 5176592ebeeb
MindTrain AGENTS.md is an instructions file published in the GitHub repository shigella520/MindTrain (120 stars, last pushed 6d ago), licensed MIT. It adds 559 tokens to every session, about $0.0028 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 instructions, from other repositories
senmu-buildos AGENTS.md
AGENTS.md instructions for SenMuShare/senmu-buildos, a project described as: AI coding governance and engineering coach for Codex, Claude Code & Doubao—improve code quality, reuse frameworks/components, reduce unnecessary code/context, and make requirements, Git, testing & releases verifiable.
supergraph AGENTS.md
Instructions for datit309/supergraph, covering supergraph — mandatory workflows, skills, auto language detection, tiered workflow — pick the right tier first and full pipeline (tier 3).
codex-token-meter AGENTS.md
Instructions for prefect12/codex-token-meter, covering ai token meter development guide, project shape, build and verification, git workflow and ai token meter test loop.
project-legibility AGENTS.md
Instructions for perhapsspy/project-legibility, covering agents.md and 검증.
lihongwei-cn CLAUDE.md
Instructions for LiHongwei-cn/lihongwei-cn, covering claude.md — claude code 项目记忆, 用户身份, 回复风格（红线）, 网站 and 技术栈.
is-ai-native copilot-instructions.md
Instructions for webmaxru/is-ai-native, a project described as: Audit public GitHub repositories for AI coding readiness across Copilot, Claude Code, and Codex.