Metadata-Version: 2.4
Name: provio
Version: 0.1.5
Classifier: Development Status :: 3 - Alpha
Classifier: Environment :: Console
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Rust
Classifier: Topic :: Security
Summary: A safety floor your AI agents can't get under: one policy for Claude Code, Codex, Gemini CLI, Cursor, Windsurf and MCP, judged by what each call will run, with a tamper-evident, signed ledger.
Keywords: ai-agents,authorization,policy,audit,mcp,claude-code
Author-email: Bhaskar Gurram <gurrambhaskar.ai@gmail.com>
License-Expression: Apache-2.0
Requires-Python: >=3.8
Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
Project-URL: Homepage, https://getprovio.vercel.app
Project-URL: Issues, https://github.com/writ-agent/provio/issues
Project-URL: Source, https://github.com/writ-agent/provio

# provio

**A safety floor your AI agents can't get under.** Every tool call an AI agent makes is checked
against one policy file (`provio.yaml`: allow / deny / ask / redact) before it
runs, and written to a hash-chained, tamper-evident ledger after.

This package installs the prebuilt `provio` command-line tool:

```bash
pip install provio      # or: npm install -g provio
provio --version
provio integrate claude-code   # govern every Claude Code tool call
provio run -- claude           # or launch an agent inside a kernel write boundary
provio verify                  # check the ledger's hash chain
```

For Python agent frameworks (LangGraph, OpenAI Agents SDK, Claude Agent SDK)
install [`provio-sdk`](https://pypi.org/project/provio-sdk/), which depends on this
package.

Documentation, policy reference and threat model:
https://github.com/writ-agent/provio

<!-- MCP Registry ownership marker -->
mcp-name: io.github.writ-agent/provio

