Metadata-Version: 2.5
Name: beadle
Version: 0.2.0
Summary: Build an AI employee without building an agent platform — deterministic checks, one-shot judgment, and a gate that keeps the model out of the loop on quiet days.
Project-URL: Homepage, https://github.com/agentropicai/beadle
Project-URL: Source, https://github.com/agentropicai/beadle
Author: Agentropic
License: MIT License
        
        Copyright (c) 2026 Agentropic
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
License-File: LICENSE
Keywords: agents,ai,automation,claude,cron,llm,monitoring
Classifier: Development Status :: 4 - Beta
Classifier: Environment :: Console
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: System :: Monitoring
Classifier: Topic :: Utilities
Requires-Python: >=3.9
Description-Content-Type: text/markdown

# Beadle

**Build an AI employee without building an agent platform.**

Most of any job is *noticing*, not thinking — scanning, comparing against a baseline, concluding
nothing changed. Noticing is constant and answerable in SQL. Judgment is rare, expensive, and the
only part a model is good at. Most agent systems fuse the two and pay judgment prices for
noticing. Beadle keeps them apart.

```
pip install beadle
beadle init my-employees
cd my-employees
./beadle doctor
./beadle run example-site-watch uptime-check
```

Every task is the same five steps:

```
GATHER   deterministic — SQL, an API call, a shell command. No model.
GATE     plain Python. Did anything happen? If not, stop here — silently.
JUDGE    one model call. Tools disabled, max-turns 1.
DELIVER  gated. Silence is the correct output most days.
RECORD   journal, archive.
```

There is no agent loop. On the production fleet this was extracted from — 21 live tasks —
10 never call a model at all, 18 can exit before the model is invoked, and the busiest task runs
144 times a day for zero tokens.

Beadle is a workspace you own and edit. `beadle init` fetches it and gets out of the way. Full
docs, four working example employees, and a complete worked build (including the bug hit along
the way) are in the workspace itself.

## Or import it into a codebase you already have

If you already have a repo with scheduled scripts in it, you do not need a second workspace:

```python
from beadle import lib          # BEADLE_HOME=/path/to/your/repo
```

`BEADLE_HOME` decides where `.env`, `logs/` and employees live; `BEADLE_EMPLOYEES` overrides the
employee root on its own, so an existing layout keeps working (`lib.journal("teams/sre", ...)`)
without moving a single folder. Everything else is identical, because it is the same `lib.py`.

The workspace remains the recommended way in: reading the four example employees teaches the
shape faster than any API docs, and editing your own copy of `lib.py` is a feature, not a
violation. Import it when you have an existing fleet to fold in rather than a new one to start.

MIT · https://github.com/agentropicai/beadle
