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/harumiweb/exstruct/agents-mdgit clone --depth 1 https://github.com/harumiWeb/exstructWhat 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 | $0.02231 | $0.02231 |
| Opus 5 | $0.01115 | $0.01115 |
| Sonnet 5 | $0.00446 | $0.00446 |
| Haiku 4.5 | $0.00223 | $0.00223 |
Grade A, and why
exstruct 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 yesterday.
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 — 183 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ExStruct AI Agents Guide
0. Overview
This repository is organized around the following top-level directories:
exstruct/
|- src/ # Main library and implementation code
|- tests/ # Automated tests
|- sample/ # Sample workbooks and example inputs
|- schemas/ # JSON schemas and validation-related assets
|- scripts/ # Utility and maintenance scripts
|- benchmark/ # Benchmark code and performance measurements
|- docs/ # User-facing documentation
|- dev-docs/ # All developer-facing documentation
|- tasks/ # Temporary task notes and working files
|- drafts/ # Draft documents and work-in-progress materials
|- dist/ # Build artifacts and packaged outputs
`- site/ # Generated documentation site output
For internal development guidance, architecture notes, ADRs, specifications, and testing references, use dev-docs/ as the canonical location. Developer-facing documentation should be written there rather than scattered across the repository.
1. Workflow Design
1. Use Plan mode by default
- Always start tasks with 3 or more steps, or tasks that affect architecture, in Plan mode
- If things stop going well partway through, do not force it; stop immediately and replan
- Use Plan mode not only for implementation, but also for verification steps
- Write detailed specifications before implementation to reduce ambiguity
2. Multi-Agent Strategy
- Actively use sub-agents to keep the main context window clean
- Delegate research, investigation, and parallel analysis to sub-agents
- For complex problems, use sub-agents to apply more compute resources
- To keep execution focused, assign one task per sub-agent
- Use explorer for read-heavy codebase exploration
- Use worker for implementation and fixes
- Use reviewer for reviews
3. Self-Improvement Loop
- Whenever you receive a correction from the user, record that pattern in
tasks/lessons.md - Write rules for yourself so you do not repeat the same mistake
- Keep improving those rules thoroughly until the error rate goes down
- At the start of each session, review the lessons relevant to the project
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.
- yesterday First seen · 183 lines · 2,231 tokens per session scan A 89fbaf5071ad
exstruct AGENTS.md is an instructions file published in the GitHub repository harumiWeb/exstruct (195 stars, last pushed 2mo ago), licensed BSD-3-Clause. It adds 2,231 tokens to every session, about $0.0112 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
tablepack AGENTS.md
Instructions for kujiangmudao/tablepack, covering two user paths, multimodal requirement, environment, hard requirements (never skip) and qc checklist (every pdf).
aiworkdeck CLAUDE.md
Claude Code instructions for zeweihan/aiworkdeck, covering claude.md, 领域文档路由表（先读文档，再动代码）, 维护规则 and 全局约定.
graphiti AGENTS.md
Instructions for getzep/graphiti, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
nuwax AGENTS.md
Instructions for nuwax-ai/nuwax, covering ai agent system documentation, 系统概述, ai agent 架构, 核心组件 and ai 功能特性.
moss AGENTS.md
Instructions for usemoss/moss, covering agents.md, what this repo is, repository layout, integrations & cookbooks and framework cookbooks (examples/cookbook/).
mcp-server-excel copilot-instructions.md
Instructions for sbroenne/mcp-server-excel, covering github copilot instructions - excelmcp, repository map, environment, working rules and build and validation.