This directory is a compact golden sample of the kind of output shape spawn is designed to produce from a recurring AI-agent workflow pattern.
It is intentionally small and readable. The point is not to be a full product, but to make the generated-output contract concrete for people scanning the repository.
A repeated session-log workflow might look like this:
Read issue text -> classify the issue -> create a triage checklist -> write a short report
A generated MCP/FastMCP server should normally include:
server.py # MCP tool surface
pyproject.toml # package/runtime metadata
README.md # generated server documentation
MCP_INFO.md # MCP metadata and integration notes
tests/ # test scaffolding in full generated outputs
The sample server.py exposes three small tools:
classify_issue— classify an issue as bug, feature, security, or ops.create_triage_checklist— create a short checklist for the classification.write_triage_report— turn classification + checklist into a report.
Without an example, spawn can sound abstract: logs, patterns, scoring, generation. This folder shows the intended end state: a repeated agent workflow becomes a reusable MCP server boundary.