What's in OneTool?

Everything you need to build powerful AI agent integrations without burning your context window.

96%

Token Savings

MCP servers consume 3-30K tokens before you start. OneTool uses ~2K tokens no matter how many tool packs you load.

No context rot. No token bloat. 24x lower cost.


Your image: Token comparison chart
See the comparison →

Explicit Execution

Write Python, not tool definitions. You see exactly what runs.

__ot brave.search(q="AI")

100+ Tools

Search, web, database, files, diagrams, conversions—batteries included.

Browse tools →

MCP Server Proxy

Wrap any existing MCP server. Configure in YAML. Call it explicitly—all the goodness of OneTool, all the power of your existing servers.

Learn more →

Developer Experience

Snippets

Reusable code templates with Jinja2. Define once, invoke with $snippet_name.

Aliases

Short names for tools. ws instead of brave.web_search.

Prefixes

p instead of pattern. Any unambiguous prefix works.

Single YAML

One config file. Three-tier inheritance: bundled → global → project.

Security

AST Code Validation

All code validated before execution. Blocks exec, eval, subprocess. Warns on risky patterns. ~1ms overhead.

Security docs →

Configurable Policies

Four tiers: Allow, Ask, Warn, Block. Fine-grained fnmatch patterns.

Path Boundaries

File ops constrained to allowed directories. Symlink-safe.

Output Sanitization

Automatic protection against indirect prompt injection. External content wrapped in GUID boundaries with triggers redacted.

Built-in Tools

Web & Search

Brave Search, Google Grounded, Firecrawl scraping.

Brave →

Smart Tools

LLM-powered transformation. Pipe output through AI.

LLM tools →

Code & Docs

Context7 library docs, semantic search, ripgrep.

Code tools →

Data & Files

SQL (any DB), Excel, PDF/Word/PPT to Markdown.

DB tools →

Extensibility

Drop-in Tools

One file, one pack. No registration, no configuration. Drop a Python file, restart, call your functions.

Create tools →

Worker Isolation

Tools with deps run in isolated subprocesses via PEP 723.

Extension Repos

Build tools in separate repos. Share via tools_dir globs.

Testing & Quality

bench Harness

Real agent + MCP server testing. Define tasks in YAML, get objective metrics: tokens, costs, accuracy, timing.

Bench docs →
1K+

Tests

Smoke, unit, integration tiers. Fast CI feedback.

Runtime Stats

Track calls, success rates, costs. HTML reports.