Metadata-Version: 2.4
Name: runvouch
Version: 0.3.3
Summary: The watchdog for unattended AI agents: dead man's switch, cost cap and outcome check for cron'd LLM scripts, Claude Code routines, OpenClaw and n8n.
Author-email: RunVouch <hello@runvouch.com>
License: MIT
Project-URL: Homepage, https://runvouch.com
Project-URL: Documentation, https://runvouch.com/docs
Project-URL: Source, https://github.com/runvouch/runvouch
Keywords: ai-agents,monitoring,watchdog,cron,llm,claude-code,openclaw,observability
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: System :: Monitoring
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# runvouch

The watchdog for unattended AI agents. Alerts within minutes when a scheduled agent is
MISSED, FAILED, reported success without evidence, stuck in a retry storm, over budget,
drifting or stalled. Telegram, Slack, e-mail or webhook. Free for 3 agents.

```bash
pip install runvouch
export RUNVOUCH_KEY=rv_...        # free key at https://runvouch.com/app

# wrap any cron job / agent command
rv run nightly-report --evidence-file out/report.html -- python report.py
```

```python
import os, runvouch
with runvouch.vouch("nightly-report", evidence=lambda: {"report": os.path.exists("out.html")}) as run:
    run.tool("openai.chat", {"model": "gpt-5"}, cost=0.01)
```

Zero dependencies, fails open (monitoring can never break your job). Docs: https://runvouch.com/docs
