| license: apache-2.0 | |
| tags: | |
| - agent | |
| - human-in-the-loop | |
| - ai | |
| - mcp | |
| - pile-protocol | |
| pretty_name: Pile Protocol | |
| # Pile Protocol | |
| Open protocol for agent-to-human decision capture. | |
| When an AI agent needs a human decision, the Pile Protocol gives it a review screen and sends back structured instructions it can act on. One link. Mark it. Agent continues. | |
| ## What this repo will hold | |
| - Pile Protocol spec (versioned) | |
| - Resolution payload schema | |
| - Conformance test suite | |
| - Example resolutions across decision types | |
| ## Status | |
| **v0.0.0 placeholder.** Spec drafting in progress. Full v0.1 RFC publishes alongside the hosted product launch. | |
| ## Links | |
| - Website: https://pileless.com | |
| - Spec: https://github.com/pileless/pile-protocol | |
| - Python SDK: https://pypi.org/project/pileless/ | |
| License: Apache-2.0 | |