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/danveil/mcp-security-inspector/agents-mdgit clone --depth 1 https://github.com/danveil/mcp-security-inspectorWrote 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/danveil/mcp-security-inspector/agents-md)<a href="https://agentmods.dev/instructions/danveil/mcp-security-inspector/agents-md"><img src="https://agentmods.dev/badge/instructions/danveil/mcp-security-inspector/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 | $0.00560 | $0.00560 |
| Opus 5 | $0.00280 | $0.00280 |
| Sonnet 5 | $0.00112 | $0.00112 |
| Haiku 4.5 | $0.00056 | $0.00056 |
Grade A, and why
mcp-security-inspector 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 5d 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 — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Purpose
MCP Tool Security Inspector performs deterministic defensive static analysis of MCP tool metadata. It must never invoke scanned tools, execute metadata, download metadata-linked resources, or send catalog content to a model.
Architecture
src/mcpsec/loader.pyandresource_policy.py: bounded hostile-input handlingsrc/mcpsec/normalizer.py: NFC normalization, alias validation, and typed tool definitionssrc/mcpsec/detectors/andrules/: built-in detectors and data-only custom rulessrc/mcpsec/risk.py: deterministic, capped risk calculationfingerprint.py,baseline.py, andcompare.py: canonical hashes and driftsrc/mcpsec/retrieval.py: explicit loopback-only MCPtools/listtransportsrc/mcpsec/evaluation/: corpus validation and metricssrc/mcpsec/reporter.pyandcli.py: inert output and CLI coordination
Commands
python -m pip install -e ".[dev]"
ruff check .
ruff format --check .
mypy src
python -m pytest --cov=mcpsec --cov-report=term-missing
python -m build
python scripts/smoke_wheel.py dist/<wheel-file>.whl
If .venv\Scripts\python.exe --version fails, recreate the virtual environment before using the helper scripts.
Security invariants
- Treat names, descriptions, schemas, annotations, metadata, rules, suppressions, baselines, and server responses as hostile data.
- Static analysis remains the default; never invoke tools or start supplied commands.
- Retrieval remains opt-in and loopback-only at the HTTP transport layer; do not enable redirects or environment proxies.
- Reject inputs over resource limits and conflicting aliases; never silently truncate or discard security-significant content.
- Preserve deterministic finding order, canonical hashes, risk caps, and stable rule IDs.
- Render hostile terminal text literally and neutralize spreadsheet formulas.
- Keep custom configuration data-only; do not introduce executable expressions or user-controlled regular expressions.
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.
- 5d ago First seen · 52 lines · 560 tokens per session scan A 25c0da7d91c9
mcp-security-inspector AGENTS.md is an instructions file published in the GitHub repository danveil/mcp-security-inspector (1 stars, last pushed 7d ago), licensed MIT. It adds 560 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 2026-08-31.
Other instructions, from other repositories
world-intel-mcp CLAUDE.md
Claude Code instructions for marc-shade/world-intel-mcp, covering claude.md, what this is, commands, install and run mcp server (stdio mode).
tooltrust-scanner CLAUDE.md
Claude Code instructions for AgentSafe-AI/tooltrust-scanner: Use the /browse skill from gstack for all web browsing. Never use mcpclaude-in-chrome tools.
mcp-audit CLAUDE.md
Claude Code instructions for adudley78/mcp-audit, covering mcp-audit, what this project does, business model, tech stack and project layout.
mcp-threatintel AGENTS.md
Instructions for aplaceforallmystuff/mcp-threatintel, covering agents.md - mcp-threatintel, tech stack, architecture, development commands and environment variables.
pkgxray CLAUDE.md
Instructions for adamsjack711-ux/pkgxray, covering project, rules, skills, getting started and commands.
mitre-mcp AGENTS.md
Instructions for lidless-labs/mitre-mcp, covering repository guidance, definition of done, project shape, verification workflow and soc safety (hard prohibitions).