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/antgroup/vsag/agents-mdgit clone --depth 1 https://github.com/antgroup/vsagWrote 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/antgroup/vsag/agents-md)<a href="https://agentmods.dev/instructions/antgroup/vsag/agents-md"><img src="https://agentmods.dev/badge/instructions/antgroup/vsag/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 | $0.01436 | $0.01436 |
| Opus 5 | $0.00718 | $0.00718 |
| Sonnet 5 | $0.00287 | $0.00287 |
| Haiku 4.5 | $0.00144 | $0.00144 |
Grade A, and why
vsag 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 — 109 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Read me first. This file is the entry point for AI coding agents working on VSAG. It is intentionally short. The bullets under Hard Constraints must hold for every change; the Documentation Index points at the detailed rules for each topic.
Project Overview
VSAG is a high-performance vector indexing library for similarity search,
written primarily in C++ with Python bindings provided by pyvsag and
Node.js/TypeScript bindings provided by vsag.
Hard Constraints (must hold every change)
- Tool versions are fixed:
clang-formatandclang-tidymust be version 15 exactly. Newer versions are not acceptable; CI enforces this. - Formatting: C++ source and header files use 4-space indentation and a
100-character line limit; Markdown and other non-C++ text files have no line
width limit. Use the
.cpp(not.cc) suffix, and ensure every committed text file ends with a trailing newline. - Layout: public APIs live in
include/vsag/; implementation lives insrc/; keep code in thevsagnamespace unless the file clearly requires otherwise. - Prefer
uint64_toversize_tin code changes to avoid macOS compile issues. - Do not modify
extern/unless the task explicitly requires third-party dependency changes. - Tests: new features need tests; bug fixes need regression coverage; the
C++ library (
src/+include/) must maintain ≥ 90 % unit-test coverage. - Commits: use Conventional Commits (
feat:,fix:,docs:,chore:, …) and DCO sign-off. AI agents must NOT add their ownSigned-off-by:— only humans can certify the DCO. AI agents must also NOT invokegit commit -s, because-sappendsSigned-off-by:at the end of the message (with a blank line separating it from earlier trailers) and breaks the required ordering. Instead, write both trailers directly in the commit message body: derive the human identity fromgit config user.name/user.email, placeSigned-off-by:first, and follow it immediately (no blank line) withAssisted-by: <AgentName>:<ModelVersion>(e.g.Assisted-by: OpenCode:claude-opus-4.7). For[skip ci], place it at the start of the subject line. - PR pushes: when preparing a pull request, prefer pushing the working
branch to the contributor's fork rather than directly to the main
antgroup/vsagrepository. Push to the main repository only when explicitly required or authorized. - Pull requests need two labels: one
kind/*(kind/bug,kind/feature,kind/improvement,kind/documentation) and oneversion/*(e.g.version/1.0). Mergify enforces both. kind/bug/kind/featurePRs must link an issue in the description using a GitHub auto-closing keyword (Fixes: #N,Closes: #N,Resolves: #N; cross-repo or full issue URL also OK).kind/improvementandkind/documentationPRs are exempt. Enforced by thePR Issue Link CheckAction and a Mergify protection.- Issues do NOT carry
Signed-off-by:— DCO only applies to commits.
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 · 109 lines · 1,436 tokens per session scan A c79d4d2fc7a3
vsag AGENTS.md is an instructions file published in the GitHub repository antgroup/vsag (507 stars, last pushed yesterday), licensed Apache-2.0. It adds 1,436 tokens to every session, about $0.0072 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 2026-09-03.
Other instructions, from other repositories
Kokoro-Engine CLAUDE.md
Claude Code instructions for chyinan/Kokoro-Engine, covering claude.md, git 规范, 项目概述, 常用命令 and 安装依赖.
retry-axios GEMINI.md
Gemini CLI instructions for JustinBeckwith/retry-axios, covering project: retry-axios, project overview, building and running, dependencies and commands.
pcl copilot-instructions.md
Copilot instructions for personamanagmentlayer/pcl, covering project overview, pcl bootstrap system, core architecture principles, 1. compiler pipeline and 2. type system.
solo copilot-instructions.md
Copilot instructions for hiero-ledger/solo, covering github copilot instructions for hiero-ledger/solo and highest-frequency rules.
openmcp-client CLAUDE.md
Instructions for LSTM-Kirigaya/openmcp-client, covering claude.md, development commands, setup and installation, development and service development.
bedrock CLAUDE.md
Claude Code instructions for christopher-buss/bedrock, covering bedrock project context, what is bedrock?, architecture, architecture quick reference and testing requirements (non-negotiable).