Can I customize how findings are categorized?

Yes, you can create a .claude/tanaoroshi.json file to define specific category names and descriptions to guide the skill's organization logic.

Does it process the logs of the Tanaoroshi skill itself?

No, the skill is designed to exclude its own execution logs to ensure the knowledge base remains focused on actual development work.

How does the skill identify valuable findings?

The skill uses specific criteria to filter out general knowledge or manual-style documentation, focusing instead on undocumented behaviors, trial-and-error results, and project-specific decision-making.

What does 'Tanaoroshi' mean?

Tanaoroshi is a Japanese term for 'inventorying' or 'stocktaking,' reflecting the skill's purpose of taking stock of the knowledge gained during your coding sessions.

Where are the extracted findings stored?

By default, findings are stored in the .claude/findings/ directory as Markdown files, organized into subdirectories by category with a central index.md file.

Tanaoroshi Knowledge Extractor

Name: Tanaoroshi Knowledge Extractor
Author: chatii

bychatii

0•

Learning & Documentation

Extracts valuable technical insights and architectural decisions from session logs into an organized project knowledge base.

Tanaoroshi is a specialized documentation skill that transforms ephemeral AI chat sessions into persistent, searchable project knowledge. By scanning session logs, it identifies non-obvious discoveries, troubleshooting steps, and technical decisions that aren't found in standard manuals. It automatically categorizes these findings, updates a centralized index, and generates project rules to ensure future AI interactions benefit from previously acquired context, effectively preventing the loss of institutional knowledge during the development lifecycle.

Key Features

01Automated extraction of project-specific technical discoveries and solutions

02Automatic generation of reference rules to enhance future Claude sessions

03Persistent Markdown-based knowledge base with centralized indexing

04Smart categorization using custom JSON configuration or automatic tag detection

050 GitHub stars

06Resource-efficient log processing using native CLI tools like jq

Use Cases

01Tracking architectural trade-offs and technical decisions made during collaborative coding sessions

02Building a project-specific 'lessons learned' database for smoother team onboarding

03Documenting complex bug fixes and environment-specific workarounds for future reference

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add chatii/claude-code-finding-tanaoroshi tanaoroshi

For use in Claude.ai and ChatGPT

Download Skill