memento-summary

You can create structured git notes from an AI coding session by consolidating decisions, problems, and context into a seven-section summary stored on refs/notes/commits. This captures verifiable evidence for downstream refinement pipelines.

Tracking AI coding decisions in git involves writing structured session summaries as git notes containing decisions, constraints, and open questions. This enforces a deterministic seven-section format with YAML frontmatter for reliable downstream loading.

Yes, you can generate session summaries automatically on git commit hooks. Applying the summarization process post-commit captures decisions and context immediately, writing the structured evidence directly to refs/notes/commits.

Writing decisions to refs/notes/commits works for knowledge distillation by attaching structured YAML frontmatter summaries to specific commits. This provides deterministic, verifiable evidence that a refinement pipeline can load and process downstream.

A structured AI coding session summary requires seven specific sections: Decisions Made, Problems Encountered, Constraints Identified, Open Questions, Context, Recorded Decisions, and Vault Entries Referenced.

No external dependencies are required to write structured git notes for traceability. The process operates independently to consolidate session context and store it within the existing git notes infrastructure.