ce-compound

To document solved problems, ce-compound analyzes conversation history, memory notes, and code context to distill a solution into a structured Markdown document with frontmatter for team knowledge discovery.

Generating structured learning documents with frontmatter is done by assembling analyzed session history and code context into a single Markdown file under docs/solutions, validating the frontmatter against a canonical schema for searchability.

The best way to classify a resolved bug as a knowledge document is using a workflow that categorizes the learning as either a bug or knowledge, then assigns a category and proposes a filename and frontmatter for discovery.

Yes, ce-compound works with subagents by coordinating a Context Analyzer, Solution Extractor, and Related Docs Finder in a parallel research workflow to extract solutions and produce a cohesive learning document.

Yes, you can integrate memory notes and code context to surface prior work and cross-session patterns through a Related Docs Finder, improving relevance and preventing duplicate knowledge documentation.

A limitation of automating knowledge management with subagents is that the workflow depends on available conversation history and memory notes to accurately classify learnings and validate frontmatter against the canonical schema.