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/airshelf/mcpfs/claude-mdgit clone --depth 1 https://github.com/airshelf/mcpfsWrote 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/airshelf/mcpfs/claude-md)<a href="https://agentmods.dev/instructions/airshelf/mcpfs/claude-md"><img src="https://agentmods.dev/badge/instructions/airshelf/mcpfs/claude-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 | $0.00833 | $0.00833 |
| Opus 5 | $0.00417 | $0.00417 |
| Sonnet 5 | $0.00167 | $0.00167 |
| Haiku 4.5 | $0.00083 | $0.00083 |
Grade B, and why
mcpfs CLAUDE.md scanned grade B with 1 finding 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
4. `~/.claude/settings.json` → `enabledPlugins` — enabled plugins (cache scan) How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
mcpfs
Mount any MCP server as a FUSE filesystem. Classifies tools into reads (files) and writes (CLI).
Quick start
go build -o mcpfs ./cmd/mcpfs # build
go test ./... # test
go vet ./... # lint
Architecture
pkg/mcpclient/— MCP JSON-RPC client (stdio + HTTP transports)pkg/mcptool/— Tool schema → CLI bridge. Parses JSON Schema into flags, dispatches calls.internal/fuse/— FUSE filesystem. Maps resources → files, templates → dirs.internal/toolfs/— Tool classification (list/get/create/update/delete/search/query)internal/config/— servers.json parser with env interpolationcmd/mcpfs/— CLI: mount, auto-discover, tool proxy, unmount
How it works
mcpfs connects to any MCP server, classifies its tools, and exposes reads as files:
list_*,get_all_*, no required params → file (dashboards.json)get_*,retrieve_*, has required params → directory (lookup by ID)create_*,update_*,delete_*→ CLI only (mcpfs tool)search_*,query_*→ CLI only (mcpfs tool)
Resources (if the server has them) are also mounted as files.
Commands
mcpfs auto # discover Claude Code plugins, mount to .mcpfs/
mcpfs auto --json # dry run — show discovered config
mcpfs auto --mount /custom/dir # mount to custom dir
mcpfs .mcpfs/posthog --http <url> # mount single HTTP server
mcpfs .mcpfs/stripe -- npx -y @stripe/mcp # mount single stdio server
mcpfs tool posthog # list tools for a server
mcpfs tool posthog create-flag --key x # call a write tool
mcpfs -u .mcpfs/posthog # unmount
Project-local mounts
mcpfs auto mounts to .mcpfs/ in the current directory (project-local).
Reads .env.local and .env from cwd for project-specific credentials, then falls back to ~/.config/mcpfs/env.
Different projects can have different API keys for the same services.
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 · 80 lines · 833 tokens per session scan B dec8dde3058f
mcpfs CLAUDE.md is an instructions file published in the GitHub repository airshelf/mcpfs (4 stars, last pushed 6mo ago), licensed MIT. It adds 833 tokens to every session, about $0.0042 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
dev3000 AGENTS.md
Instructions for vercel-labs/dev3000, covering agents.md, runtime, local ui, production workflows, browser tools and development rules.
agent-inspect AGENTS.md
AGENTS.md instructions for rajudandigam/agent-inspect, covering agentinspect ai maintainer instructions, product boundary, source of truth, public-copy rule and start every task.
jenkins-cli CLAUDE.md
Claude Code instructions for avivsinai/jenkins-cli, covering claude.md, build & test commands, run a single test, skip e2e tests during unit testing and e2e with colima on macos (if docker is unreachable).
patchloom AGENTS.md
AGENTS.md instructions for patchloom/patchloom, covering agents.md, project overview, dev commands, git hygiene and release prs (release-please).
claude-crusts CLAUDE.md
Claude Code instructions for Abinesh-L/claude-crusts, covering identity, cli commands, rules, format and architecture.
goal-flight AGENTS.md
AGENTS.md instructions for simonrowland/goal-flight, covering agent notes, goal flight routing, project commands (this repo), git workflow (this repo) and git-visible trigger hygiene.