Skip to content

Your First Week with StudyLoop

A minimal, repeatable path for getting value from StudyLoop without configuring every agent or LLM provider on day one. The core workflow is live Socratic study; flashcards, quizzes, and session export support that loop.

What you need

  • Python 3.12+, uv, and tmux 3.1+ (for studyloop study)
  • One supported AI coding agent on your PATH: Kiro CLI, Codex, Claude Code, OpenCode, or pi. Kiro is the documented first-session path.
  • Optional: Obsidian vault for study notes; optional: Ollama or AWS Bedrock for local/cloud card generation

Install from source (see Setup Guide):

git clone https://github.com/NetDevAutomate/StudyLoop.git studyloop
cd studyloop
./scripts/install.sh
studyloop setup
studyloop doctor --fix

Day 1 — Install and one live session

  1. Run studyloop doctor --fix until core checks pass.
  2. Start a short study session (pick one topic you care about):
studyloop study "Python decorators" --energy 6

Or use the web picker:

studyloop web

Open the Study Session tab, choose a topic and agent, and start.

  1. At the end of the session, run studyloop study --end or end from the web UI.

  2. Check context for next time:

studyloop resume

Day 2 — Let StudyLoop choose one next action

  1. Ask for one useful next action instead of picking from the whole vault:
studyloop now --energy medium --time 20
  1. Open the source note it points at, or pick one note from your current course:
studyloop chat-note ~/Obsidian/Personal/Study/Python/decorators.md --mode recall
  1. End with a small evidence record. Use the command printed by studyloop now or chat-note, usually studyloop progress or studyloop teachback.

Day 3 — Session memory and wins

  1. Export recent agent sessions into the shared database (from any tool you already use):
session-export
session-query list --since last-7-days

--since takes YYYY-MM-DD, last-week, last-month, or last-N-days. Shorthand like 7d is rejected.

  1. Surface progress (useful for AuDHD “nothing counts” moments):
studyloop wins
studyloop streaks
studyloop review
studyloop recap today
  1. If you use Kiro and want struggle signals in progress tracking:
studyloop extract-struggles --incremental --dry-run --model <bedrock-model-id>

--model is required — it names the Bedrock model that reads your sessions. Set STUDYLOOP_EXTRACTOR_MODEL once and you can leave the flag off. This step needs AWS credentials; skip it if you have none configured.

Day 4 — Review support and source notes

If you have markdown notes or a course folder:

studyloop content discover
studyloop content generate-cards ~/path/to/course-notes --course my-course
studyloop content generate-practice ~/path/to/course-notes --course my-course
studyloop web

Use Flashcards or Quizzes in the web app. In the Course Explorer, open a lesson and use Discuss to copy a Socratic prompt from the current note; use Struggling? when the lesson should feed future repair work. You do not need NotebookLM for this path.

Practice tasks stay in the terminal. The web app has no exercises panel yet, so review the JSON that generate-practice wrote with studyloop practice verify TASKS.json --task 1 --notes "what passed".

For generation, configure one provider in the web Settings → LLM Providers panel (Ollama locally or Bedrock/OpenAI if you already have keys). See Content Pipeline.

Day 5 — Agents and backlog

  1. Install agent definitions for your tool:
studyloop install agents
  1. Park tangents instead of rabbit-holing:
studyloop park "How does asyncio compare to threads?" -t python
studyloop backlog list
  1. Check that your chosen mentor runs studyloop resume and studyloop review at session start. Reinstall the shipped agent definitions with studyloop install agents if it does not.

Use the mastery commands once you have a few progress records:

studyloop mastery graph --topic python
studyloop mastery weak-links --topic python
studyloop review --interleave adaptive --energy low

The graph is Mermaid by default, so it can be pasted into Obsidian.

Day 7 — Hardening habits

  • Run studyloop clean --dry-run if tmux study sessions ever feel “stuck”.
  • Set studyloop backup before big config experiments.
  • Skim Web UI Guide for LAN mode (studyloop web --lan) if you review from a tablet or another laptop on the same trusted network. Phone screens are not supported.

What to defer

Later Why
Every LLM provider in Settings One working backend is enough for generation
NotebookLM sync commands Legacy optional path; not required for core study
session-export --obsidian Opt-in vault mirror; enable when you want Dataview notes
All five agent CLIs Start with the one you already use daily

Where to go next