What are the prerequisites for using this skill?

You must have a .rivet/systems.yaml file already initialized in your project, typically created by running a project scan first.

What triggers a 'harvest' in the chat?

You can invoke the skill by saying 'harvest', 'capture decisions', or 'save what we discussed', or by using the /rivet-harvest command.

Where does Rivet Harvest save its data?

The skill saves all confirmed items into a .rivet/systems.yaml file located in your project's root directory.

What is Rivet Harvest?

Rivet Harvest is a Claude Code skill that scans your chat history to identify, extract, and save architectural decisions, requirements, and terminology.

Do I have to save every decision the AI finds?

No. The skill provides an interactive selection process where you can review, edit, or reject potential items before they are written to your documentation.

Rivet Harvest

Name: Rivet Harvest
Author: ratacat

byratacat

Learning & Documentation

Extracts and persists architectural decisions, requirements, and terminology from conversation transcripts into structured project documentation.

About

Rivet Harvest solves the problem of ephemeral knowledge by scanning Claude Code conversation history to identify and extract critical technical decisions, system requirements, and project-specific terminology. Instead of letting valuable architectural insights vanish after a session, this skill identifies patterns in your chat history and prompts you to save them into a structured .rivet/systems.yaml file. This ensures that the evolving logic, constraints, and vocabulary of your codebase remain documented, searchable, and accessible for future development cycles.

Key Features

Automated transcript scanning to identify project decisions and requirements
Interactive selection interface to review and approve extracted items
Project-level glossary management for team-specific terminology
Pattern-based extraction of keywords like 'must', 'should', and 'decided'
0 GitHub stars
Direct integration with .rivet/systems.yaml for structured persistence

Use Cases

Persisting architectural choices made during brainstorming sessions with Claude
Maintaining a living glossary of project-specific terms and definitions
Automatically updating project requirements as they are defined in natural language

About

Key Features

Automated transcript scanning to identify project decisions and requirements
Interactive selection interface to review and approve extracted items
Project-level glossary management for team-specific terminology
Pattern-based extraction of keywords like 'must', 'should', and 'decided'
0 GitHub stars
Direct integration with .rivet/systems.yaml for structured persistence

Use Cases

Persisting architectural choices made during brainstorming sessions with Claude
Maintaining a living glossary of project-specific terms and definitions
Automatically updating project requirements as they are defined in natural language