observe

Extract typed observations from session transcripts and score them by importance, confidence, and surprise.

1|Updated Mar 6, 2026

Search Tags:#automation #memory #nlp #transcripts #log-analysis #observations

Ready to run

npx skills add https://github.com/adiman9/mnemos --skill observe-adiman9

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: observe
Source: https://github.com/adiman9/mnemos/tree/main/core/skills/observe
Command: npx skills add https://github.com/adiman9/mnemos --skill observe-adiman9

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Extract typed observations from session transcripts and convert them into structured notes for memory and daily review, enabling faster analysis and more actionable recall.

Core Features & Use Cases

Incrementally process transcripts stored under vault_path/memory/sessions.
Score observations by importance, confidence, and surprise, and route to daily logs and MEMORY.md.
Support cron-friendly incremental execution with cursor tracking and optional scope filters.

Quick Start

Process new transcripts from the vault and emit typed observations to the daily logs.

Frequently Asked Questions about observe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I extract structured observations from session transcripts?▼

To extract structured observations from session transcripts, this Skill processes files stored under vault_path/memory/sessions, scoring them by importance, confidence, and surprise, and routes the typed notes to daily logs and MEMORY.md.

Can I automate transcript log analysis to run incrementally on new sessions?▼

Yes, transcript log analysis supports cron-friendly incremental execution with cursor tracking, allowing you to process only new session transcripts and append observations to daily logs without reprocessing previous files.

How does NLP scoring work for typed observations in daily logs?▼

NLP scoring for typed observations evaluates each extracted note by assigning values for importance, confidence, and surprise, providing a quantifiable structure for prioritizing transcript insights in your memory logs.

What is the best way to organize session transcripts for automated observation extraction?▼

Organize session transcripts under the vault_path/memory/sessions directory to enable automated observation extraction, which routes the structured output directly to vault_path/memory/daily logs and MEMORY.md.

Does this transcript observation tool prevent fabrication of notes not present in the source text?▼

Yes, the transcript observation tool enforces a strict, self-contained output format that prevents fabrication, ensuring all extracted notes and scored observations are grounded strictly in the source session transcripts.

Can I apply scope filters to limit which session transcripts get processed?▼

Yes, you can apply optional scope filters during transcript processing to limit observation extraction to specific sessions, ensuring only targeted data is routed to your daily logs and MEMORY.md.