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/agentverus/agentverus-scanner/agents-mdgit clone --depth 1 https://github.com/agentverus/agentverus-scannerWhat 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 | $0.02548 | $0.02548 |
| Opus 5 | $0.01274 | $0.01274 |
| Sonnet 5 | $0.00510 | $0.00510 |
| Haiku 4.5 | $0.00255 | $0.00255 |
Grade A, and why
agentverus-scanner 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 — 260 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — AgentVerus Scanner
Project Overview
agentverus-scanner is the open-source (MIT) scanning engine that powers AgentVerus. It analyzes AI agent skill files (SKILL.md and variants) for security and behavioral risks, produces trust reports with scores 0–100, and outputs SARIF for CI integration.
Published on npm: agentverus-scanner
This is the core IP — the analysis engine. The web app (agentverus-web) is a separate repo that wraps this library with an API, database, and UI.
Tech Stack
| Layer | Technology | Notes |
|---|---|---|
| Language | TypeScript 5.7+ | Strict mode. ESM only. |
| Runtime | Node.js 22+ | Zero runtime dependencies (only fflate for zip) |
| Testing | Vitest | Unit + integration. Coverage target: 80%+ |
| Linting | Biome | Tab indent. Double quotes. Line width 100. |
| Package Manager | pnpm | Workspace monorepo (pnpm-workspace.yaml) |
| CI Action | GitHub Actions (composite) | actions/scan-skill/action.yml |
Zero-dependency philosophy
The scanner has exactly one runtime dependency (fflate for zip decompression). Everything else is built from scratch. No OpenAI SDK, no heavy parsers, no framework bloat. Keep it that way.
Architecture
src/
├── scanner/ # Core scanning engine (npm package entry)
│ ├── index.ts # Orchestrator: scanSkill(), scanSkillFromUrl()
│ ├── parser.ts # Multi-format SKILL.md parser
│ ├── analyzers/ # One file per analysis category
│ │ ├── permissions.ts # ASST-03, ASST-08
│ │ ├── injection.ts # ASST-01, ASST-06
│ │ ├── dependencies.ts # ASST-04
│ │ ├── behavioral.ts # ASST-07, ASST-09
│ │ ├── content.ts # ASST-02, ASST-05, ASST-10
│ │ ├── semantic.ts # LLM-assisted deep analysis (optional)
│ │ ├── context.ts # Cross-analyzer context enrichment
│ │ └── declared-match.ts # Permission declaration matching
│ ├── scoring.ts # Per-analyzer weighted score aggregation
│ ├── score-calibration.ts # Overall-score calibration + badge-tier rules
│ ├── runner.ts # Batch scan orchestration
│ ├── targets.ts # Target expansion (files, dirs, URLs, globs)
│ ├── sarif.ts # SARIF 2.1.0 output formatter
│ ├── source.ts # URL fetching with retries
│ ├── cli.ts # CLI entry point (`agentverus` command)
│ └── types.ts # All type definitions + ASST taxonomy
├── registry/ # Batch scanning + HTML report generation
│ ├── index.ts # Registry orchestrator
│ ├── cli.ts # Registry CLI subcommands
│ ├── batch-scanner.ts # Parallel batch scanning
│ ├── report-generator.ts # Markdown report output
│ ├── site-generator.ts # Static HTML site generator
│ └── types.ts # Registry-specific types
packages/
└── agentverus-scanner-mcp/ # MCP server wrapper (separate package)
actions/
└── scan-skill/ # GitHub Action (composite)
├── action.yml
└── dist/index.cjs # Bundled action entry
test/
├── scanner/ # Scanner tests (mirrors src/scanner/)
└── registry/ # Registry tests
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 · 260 lines · 2,548 tokens per session scan A 5b3ffc4c8300
agentverus-scanner AGENTS.md is an instructions file published in the GitHub repository agentverus/agentverus-scanner (8 stars, last pushed 2mo ago), licensed MIT. It adds 2,548 tokens to every session, about $0.0127 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
orbit CLAUDE.md
Instructions for schmitech/orbit, covering claude.md, project overview, commands, setup and running the server.
CORE CLAUDE.md
Instructions for DariuszNewecki/CORE, covering claude.md — core, what core is, and what you are, source layout, how to work in this repo and governed and prohibited surfaces.
dsh-auto-review AGENTS.md
Instructions for PerryLink/dsh-auto-review, covering agents.md, layout, hard rules applied here, build and docs.
orloj CLAUDE.md
Instructions for OrlojHQ/orloj, covering orloj claude instructions, critical repository rules and implementation notes.
qwed-verification copilot-instructions.md
Instructions for QWED-AI/qwed-verification: Read and follow QWEDRULES.md for every suggestion.
ab900 CLAUDE.md
Instructions for timothywarner-org/ab900, covering claude.md, repository purpose, architecture, commands and 2. non-ascii punctuation -- must return zero.