SageOx for solo players
You don't need anyone's buy-in. This cookbook walks through a week of solo SageOx usage — from first install to a personal knowledge base your AI coworkers reference automatically.
By the end, you'll have a searchable record of your decisions, your debugging sessions, and your architecture thinking. And when coworkers eventually join, everything you've built is already there.
Day 1: install and connect
You now have a team (population: you) and a connected repo. ox init sets up Claude Code integration, so every session starts with your Team Context.
Index your codebase
CodeDB gives Claude Code a semantic map of your repo — function search by meaning, git history queries, and cross-reference navigation. This alone makes coding sessions noticeably better.
Run your first context-aware session
If you have Claude Code installed, start a session:
Try asking your AI coworker about your codebase. With CodeDB indexed, it navigates with precision instead of grep-and-hope.
Day 2: import existing recordings
You have Loom walkthroughs and Cap recordings sitting in Slack threads. Import them.
Each recording gets transcribed, summarized, and committed to your Team Context. Your next Claude Code session can reference the decisions and context from those walkthroughs.
Day 3: record your first voice memo
Open the web recorder on your phone (add it to your home screen). Talk for 60 seconds about a decision you made today.
Examples worth recording:
- "Going with retry-with-backoff for the upload middleware because the S3 presigned URLs expire during long uploads and a naive retry hits the same expired URL."
- "The notification service needs to be extracted from the monolith before we can scale independently. Blocking on the event schema."
- "Found the root cause of the flaky test — the mock clock wasn't advancing during the timeout check. Fixed by injecting a real timer in test mode."
60 seconds. No editing. SageOx transcribes and commits it automatically.
Day 4–5: build the habit
Record a voice memo or short Loom every time you:
| Moment | What to capture |
|---|---|
| Make a decision | The choice and why — alternatives you rejected |
| Fix a tricky bug | Symptoms, false leads, root cause, fix |
| Design something | The constraints driving the design |
| Hit a wall | What you tried, what failed, what you need |
This is your engineering journal — except you talk instead of write, and your AI coworkers can search it.
Day 6–7: see the compounding
By now your Team Context has:
- CodeDB index of your entire repo
- Transcripts and summaries from imported recordings
- Voice memos capturing your recent decisions
Start a Claude Code session and ask about something you recorded earlier. Your AI coworker cites your own words back to you — the decision, the reasoning, the context.
This is the compounding loop. Every recording makes future sessions smarter. Every session can produce new insights worth recording.
When coworkers join
The transition from solo to team is one command:
Everything you've built — imported recordings, voice memos, session transcripts — is immediately available to their AI coworkers too. They run ox code index to set up CodeDB for their own sessions. No migration, no setup, no "let me walk you through everything." The context is already there.
What's next
- SageOx + Loom — deep dive on importing and recording walkthroughs
- SageOx + Claude Code — advanced patterns for coding sessions
- SageOx + Slack via OpenClaw — capture team decisions from Slack when your team grows