OneTool MCP

🧿 One MCP for developers - no tool tax, no context rot.
250+ tools your agent calls as Python code: search, docs, files, databases, diagrams, vision, memory - plus a proxy for every MCP server you already use.

Works with Claude Code, Cursor, Codex - any MCP client

---

The Problem

Every MCP server re-sends its tool definitions on every request: 3K-30K tokens each. Connect 5 servers and you've burned 55K tokens before the conversation starts. Connect 10+ and you're at 100K.

The math is brutal: Claude Opus 4.5 at $5/M input tokens, 20 days × 10 conversations × 10 messages × 3K tokens = $30/month per MCP server - even if you never use the tools.

And then there's context rot - your AI literally gets dumber as you add more tools (Chroma Research, 2025).

The Solution

OneTool is one MCP server that exposes tools as a Python API. Instead of reading tool definitions, your agent writes code:

__onetool brave.search(query="react 19 server components")

Configure one MCP server. Use unlimited tools - ~2K tokens no matter how many you add.

"Agents scale better by writing code to call tools instead. This reduces the token usage from 150,000 tokens to 2,000 tokens...a cost saving of 98.7%"

— Anthropic Engineering

97% fewer tokens. 30× lower cost. No context rot. (Measured - 47,660 → 1,131 input tokens against 18 MCP servers.)

📖 Read the full story

---

Code Is the Interface

Because tools are Python functions, your agent does things tool-call JSON can't: batch, chain, loop, compose.

__onetool
page = webfetch.fetch(url="https://fastmcp.dev/changelog", output_format="markdown")
notes = ot_llm.transform(data=page, prompt="Summarise the breaking changes")
mem.write(topic="deps/fastmcp", content=notes)

Three packs, one request. Intermediate results flow between tools as variables - the page body never touches your context window, and the summarising runs on a cheap model instead of your expensive coding agent.

Every call is explicit and reviewable - __onetool brave.search(query="...") shows you exactly what runs. No tool-selection guessing.

And the runtime is built for how agents actually type:

• mem.search(q="auth") works - any unambiguous parameter prefix resolves (q= → query=)
• wb.draw(...) works - packs have short aliases (wb, ctx, img)
• github.listRepositories() works on proxied servers - snake/camel/Pascal all resolve
• A typo'd tool gets a did-you-mean, a disconnected server names the command that fixes it
• Oversized results come back as a searchable handle instead of flooding the window

---

Install

Bootstrap (installs uv if missing, installs OneTool, initialises config, prints MCP config):

curl -LsSf https://onetool.beycom.online/install.sh | sh          # macOS / Linux
irm https://onetool.beycom.online/install.ps1 | iex               # Windows (PowerShell)

Or install manually with uv:

uv tool install 'onetool-mcp[all]'   # everything
onetool init --config ~/.onetool

Then print ready-to-paste MCP client config with resolved absolute paths and add it to your client (claude-code, claude-desktop, cursor, or vscode):

onetool init mcp-config --client claude-code   # or omit --client for all four

That's it. All 250+ tools work out of the box.

Verify: onetool init validate --config ~/.onetool/onetool.yaml

Install the ot-ref skill into your agent with vercel-labs/skills - it teaches the call conventions and ships a greppable index of every tool signature:

npx skills add https://github.com/beycom/onetool-mcp --skill ot-ref --agent claude

📖 Full installation guide

---

What's Inside

Search & docs	Brave, Google-grounded, and Tavily search (each with batch + answer modes), Context7 library docs, web fetch with extraction controls
Files & data	File ops with path boundaries, full Excel control, SQL databases, PDF/Word/PowerPoint → Markdown, ripgrep, package versions
Context economy	ctx handles for large outputs, partial file reads (toc/slice), image vision on a dedicated cheap model (zero host tokens), LLM delegation (10× savings)
Persistent state	mem memory with semantic + keyword search, history and rollback; knowledge RAG bases with AI enrichment; localhist Git-backed project snapshots
Visual	Live Excalidraw whiteboard with a Mermaid-compatible DSL and offline auto-layout, Mermaid/PlantUML/D2 diagrams, architecture models → draw.io-editable SVG
Runtime	MCP server proxy with runtime enable/disable/restart, direct CLI/API into the running process, ot-ref agent skill, in-conversation tool forging
Trust	age-encrypted secrets backed by your OS keychain, AST validation, path boundaries, output sanitisation, runtime stats with estimated savings

---

Tools

28 packs, 253 tools ready to use (console in beta):

Pack	Tools	Extra	Description
arch	generate, validate, bundle_solution, …	[dev]	Architecture models → draw.io-editable SVG
brave	search, news, image, video, search_batch	[util]	Brave web search
chrome_util	highlight_element, guide_user, …	[dev]	Browser annotations (Chrome DevTools)
console (beta)	show, display, list, read, clear		Messages to the upcoming onetool-console app
context7	search, doc	[dev]	Library documentation
convert	pdf, word, powerpoint, excel, auto	[util]	Documents → Markdown
db	query, schema, tables, sample	[dev]	SQL databases
diagram	render_diagram, batch_render, get_template, …	[dev]	Mermaid / PlantUML / D2 via Kroki
excel	read, write, formula, create_table, … (24 tools)	[util]	Full Excel control
file	read, write, edit, grep, slice, toc, … (16 tools)	[util]	File ops with path boundaries
ground	search, dev, docs, reddit, search_batch	[util]	Google-grounded search with sources
knowledge	search, ask, write, related, … (15 tools)	[util]	RAG knowledge bases (hybrid search)
localhist	save, diff, restore, autosave_start, … (15 tools)	[dev]	Git-backed local history snapshots
mem	write, search, ask, history, rollback, … (31 tools)	[util]	Persistent memory with semantic search
ot	help, tools, stats, status, result, … (18 tools)		Introspection and management
ot_context (ctx)	write, read, grep, slice, toc, ask, … (13 tools)		Smart context store for large outputs
ot_forge	create_ext, validate_ext		Scaffold new tool packs
ot_image (img)	load, ask, clip_ask, summary, … (9 tools)		Image vision via a dedicated model
ot_llm	transform, transform_file		LLM-powered transforms
ot_secrets	set, encrypt, audit, rotate, … (8 tools)		Encrypted secrets management
ot_servers	enable, disable, restart, status		Runtime control of proxied servers
ot_timer	start, stop, elapsed, list, clear		Named timers
package	pypi, npm, version, audit, models	[dev]	Package versions and staleness
play_util	highlight_element, guide_user, …	[dev]	Browser annotations (Playwright)
ripgrep	search, count, files, types	[dev]	Fast code search
tavily	search, research, extract, search_batch, …	[util]	AI-native search and extraction
webfetch	fetch, fetch_batch	[dev]	Web content extraction
whiteboard (wb)	open, draw, layout, screenshot, … (22 tools)	[util]	Live Excalidraw canvas

📖 Complete tools reference — every signature, generated from source

---

MCP Server Proxy

Keep the MCP servers you already use. Wrap them in YAML and call them explicitly - as Python namespaces, without their tool tax:

# .onetool/onetool.yaml
servers:
  local_tools:
    type: stdio
    command: npx
    args: ["-y", "some-mcp-server@latest"]
  private_api:
    type: http
    url: ${PRIVATE_MCP_URL}
    auth:
      type: bearer
      token: ${PRIVATE_MCP_TOKEN}

__onetool private_api.read_resource(path="README.md")

Proxied servers can be enabled, disabled, and restarted mid-conversation with ot_servers - no client restart.

📖 Configuration guide

---

Secrets You Can Commit

onetool init walks you through encrypted secrets: values in secrets.yaml are age-encrypted, the private key lives in your OS keychain, and decryption happens transparently at load.

# secrets.yaml - safe to inspect, safe to commit
brave_api_key: age1enc:YWdlLWVuY3J5cHRpb24ub3JnL3YxCi0+IFgyNT...

📖 Security guide

---

Use from the CLI

Works as an MCP server and as a direct CLI bridge into the same running process - loaded config, secrets, and proxy connections stay warm. Useful for agent harnesses, scripts, and automation:

# Recommended local MCP root mode: stdio
onetool serve --config .onetool/onetool.yaml

# URL-based MCP root mode for containerized clients
onetool serve --transport http --config .onetool/onetool.yaml --host 127.0.0.1 --port 8767 --path /mcp

# Enable the MCP-owned direct API in onetool.yaml:
# direct.host.enabled: true

# Start OneTool as MCP, then use the port printed in startup logs.
onetool direct run --port 8765 "ot.packs()" --format json | jq '.[0].name'
onetool direct run --port 8765 "brave.search(query='latest AI news')" --format raw

📖 Direct usage guide

---

Extending

Drop a Python file, get a pack. No registration, no config:

# .onetool/tools/wiki.py
pack = "wiki"

def summary(*, title: str) -> str:
    """Get Wikipedia article summary."""
    import httpx
    url = f"https://en.wikipedia.org/api/rest_v1/page/summary/{title}"
    return httpx.get(url).json().get("extract", "Not found")

__onetool wiki.summary(title="Python_(programming_language)")

📖 Creating tools guide

---

Documentation

• Quickstart - 30 seconds to first tool call
• Installation - All platforms
• Configuration - YAML schema
• Tools Reference - All 253 tools
• Security - The layered security model
• Extending - Build your own
• Dev Docs - Internal developer documentation
• Specifications - OpenSpec specifications index

---

References

• Code Execution with MCP - Anthropic Engineering
• Context Rot - Chroma Research

---

Telemetry

OneTool sends anonymous startup pings (event type, version, OS). No personal data. Opt out: export DO_NOT_TRACK=1 or set telemetry.enabled: false in onetool.yaml. Details

---

Issues

Check for existing issues first:

• Browse the tracker: github.com/beycom/onetool-mcp/issues
• Search with GitHub syntax: is:issue repo:beycom/onetool-mcp <keyword>

Raise a new issue: github.com/beycom/onetool-mcp/issues/new

---

Support

If you find OneTool useful:

---

License

GPLv3