Tacit.memorize

Skill sangmin7648/tacit/skills/tacit.memorize

Harness tacit knowledge into the context of AI agent with on device always-on transcription

Install

npx -y skills add sangmin7648/tacit --skill tacit.memorize

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
9 stars9 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Summarize the current session and save it to the tacit knowledge base (~/.tacit/memory/). TRIGGER when: the user invokes /tacit.memorize, wants to save the current conversation to memory, or asks to record what was discussed. DO NOT TRIGGER for: querying existing knowledge (use /tacit.knowledge instead).

SKILL.md

3.2 KB, as published. Nobody here has run it

tacit Memorize — Session → Knowledge Entry

Save the current conversation as a structured knowledge entry to ~/.tacit/memory/<category>/YYYYMMDD-HHMMSS.md.

Process

Step 1 — Analyze the conversation

Read the current conversation and extract:

title: A concise title (≤ 100 chars) capturing the main topic — write in the conversation's language
category: A single English word describing the domain (e.g. dev, meeting, learning, design, ideas, debugging) — always English
keywords: 5–10 search terms to maximize lexical recall in tacit search. Include:
- Synonyms and related concepts in the conversation's language plus English equivalents
- Abbreviations and alternative phrasings someone might search for
- Specific names: tools, methods, people, places mentioned
summary: Bullet list of key insights, decisions, and learnings — write in the conversation's language
content: Detailed notes — background, context, rationale, specifics — write in the conversation's language

If the user provided a hint with the command (e.g. /tacit.memorize skill development), use it to guide the title and category.

Step 2 — Determine the file path

Run the following to get a timestamp:

date +%Y%m%d-%H%M%S

Also get the current timezone offset for created_at:

date +%z

Build the paths:

CATEGORY_DIR = ~/.tacit/memory/<category>
FILE_PATH = ~/.tacit/memory/<category>/YYYYMMDD-HHMMSS.md

Create the directory:

mkdir -p ~/.tacit/memory/<category>

Step 3 — Write the file

Write the file using the Write tool at the exact path from Step 2. Use this format:

---
title: "<title>"
category: "<category>"
created_at: "<YYYY-MM-DDTHH:MM:SS+HH:MM>"
keywords: ["<keyword1>", "<keyword2>", ...]
---

<summary — bullet list of key points>

---

<content — detailed notes, background, context>

Format rules:

title and category must be quoted strings in the YAML frontmatter
created_at must be ISO 8601 with timezone offset (e.g. "2026-04-02T15:04:05+09:00")
keywords must be a JSON-style inline array of quoted strings on one line
Summary and content are plain Markdown (no YAML)
Separate frontmatter from summary with ---\n\n (blank line after)
Separate summary from content with \n\n---\n\n (blank lines around)

Step 4 — Report

Print the saved file path so the user can verify:

Saved to: ~/.tacit/memory/<category>/YYYYMMDD-HHMMSS.md

Important

Write to the absolute path (expand ~ to the actual home directory path)
Do NOT fabricate or embellish — only record what actually appeared in the conversation
Summary = concise takeaways; Content = full detail
Categories are single-level, no slashes (e.g. dev not dev/backend)
The entry will automatically appear in tacit list and tacit search — no additional steps needed