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/lianghsun/open-sheet/agents-mdgit clone --depth 1 https://github.com/lianghsun/open-sheetWrote 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/lianghsun/open-sheet/agents-md)<a href="https://agentmods.dev/instructions/lianghsun/open-sheet/agents-md"><img src="https://agentmods.dev/badge/instructions/lianghsun/open-sheet/agents-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 | $0.01070 | $0.01070 |
| Opus 5 | $0.00535 | $0.00535 |
| Sonnet 5 | $0.00214 | $0.00214 |
| Haiku 4.5 | $0.00107 | $0.00107 |
Grade A, and why
open-sheet 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 4d 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 — 88 lines — stays where its author put it; the contents beside it link to each section on GitHub.
open-sheet — repository guide
This file describes the framework repo. If you are authoring workbooks in a scaffolded workspace, you want the sheet-authoring skill instead.
Repo layout
pnpm + Turbo monorepo.
| Path | Package | Purpose |
|---|---|---|
packages/core |
@open-sheet/core |
Compiler, placement, references, formula engine, style bridge, export, viewer, Vite plugin, open-sheet CLI |
packages/cli |
@open-sheet/cli |
Scaffolder and project template |
packages/mcp |
@open-sheet/mcp |
MCP server over Streamable HTTP |
apps/demo |
private | Local framework consumer, dogfood target |
Shared config at the root: biome.json, turbo.json, pnpm-workspace.yaml, tsconfig.base.json.
Commands
pnpm dev # run the demo against local core
pnpm build # build all packages
pnpm typecheck # tsc across the monorepo
pnpm check # biome: format, lint, organize imports
pnpm check:fix # auto-fix
pnpm test # vitest
Filter to one package: pnpm core <script>, pnpm cli <script>, pnpm mcp <script>.
The one invariant
No A1 address is ever authored by hand — not in apps/demo, not in fixtures, not in tests that represent user code.
The whole premise of open-sheet is that the framework owns every coordinate. A test fixture containing =SUM(B2:B13) written by a human is a fixture that stops testing the thing that matters. Tests may assert on A1 output; they must not author it.
The one exception is raw(), the deliberate escape hatch. It is expected to show #NOT_EVALUATED in the viewer.
Core pipeline
Understand this before touching packages/core:
sheets/<id>/index.tsx
→ (1) evaluate JSX via our own jsx-runtime (no react-dom, no DOM at all)
→ Block tree
→ (2) measure each block reports its intrinsic { rows, cols }
→ (3) place assign every block an origin; collisions are a bug
→ Grid model + anchor registry
→ (4) resolve Refs → concrete A1 / defined names
→ (5) evaluate topological evaluation → the numbers the viewer shows
→ emit: viewer | .xlsx | .csv | .html | .pdf
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.
- 4d ago First seen · 88 lines · 1,070 tokens per session scan A 0f215852ce30
open-sheet AGENTS.md is an instructions file published in the GitHub repository lianghsun/open-sheet (120 stars, last pushed 8d ago), licensed MIT. It adds 1,070 tokens to every session, about $0.0053 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
excel-mcp-server AGENTS.md
Instructions for TangentDomain/excel-mcp-server, covering excelmcp, turn 起始流程, 项目事实(l0), 技术栈 and 目录结构.
excel-mcp-server CLAUDE.md
Instructions for TangentDomain/excel-mcp-server, covering excel mcp server, 技术栈, 目录结构, mcp 工具 → python api 映射 and sql 核心原则.
tablepack AGENTS.md
Instructions for kujiangmudao/tablepack, covering two user paths, multimodal requirement, environment, hard requirements (never skip) and qc checklist (every pdf).
advanced-calc-engine AGENTS.md
AGENTS.md instructions for surajpanwar/advanced-calc-engine, covering advancedcalc engine, project purpose, architecture, core rules and agent responsibilities.
exstruct CLAUDE.md
Instructions for harumiWeb/exstruct, a project described as: Conversion from Excel to structured JSON (tables, shapes, charts) for LLM/RAG pipelines, and autonomous Excel reading/writing by AI agents via CLI and MCP integration.
mcp-server-excel readme-management.instructions.md
Instructions for sbroenne/mcp-server-excel, covering readme management - quick reference, core entry documents, feature documentation, critical rules and tool & action counts must match.