Getting Started¶
This walkthrough follows Waymark's core loop end to end: capture a memory, find it again, ask a grounded question, and reflect.
1. Install¶
py -m pip install --user pipx
py -m pipx ensurepath
py -m pipx install waymark-memory
brew install pipx
pipx ensurepath
pipx install waymark-memory
python3 -m pip install --user pipx
python3 -m pipx ensurepath
python3 -m pipx install waymark-memory
The Installation & Updates guide also includes direct GitHub Release wheel commands.
2. Pick where your data lives¶
By default Waymark stores everything in a per-user home directory. For experiments, point it somewhere disposable:
$env:WAYMARK_HOME = "$PWD\.waymark-runtime"
export WAYMARK_HOME="$PWD/.waymark-runtime"
Run waymark doctor to see the resolved home, your local capability profile, and
whether Ollama is detected.
3. See today's loop¶
waymark today
This shows captures from today, reflection windows that need attention, decision review items, and copyable next commands.
4. Capture a memory¶
waymark capture --type project "Decided to build the CLI before the desktop app."
Waymark drafts a structured memory card — a title, a summary, a type, and tags — and saves it. Want to see the draft before committing?
waymark capture --preview "Draft this card without saving it yet."
The guided interface (waymark) uses a two-step Draft → Save / Edit / Discard
flow for the same capture.
5. Browse the timeline¶
waymark timeline
Each row includes a memory ID. Open one in detail:
waymark memory show 1
6. Ask a grounded question¶
waymark ask "CLI first"
Answers are composed only from your saved memories and always cite the
sources they used. If nothing matches, Waymark says so instead of inventing an
answer. Add --semantic to include vector matches once you've generated
embeddings (see Local AI).
7. Reflect¶
waymark reflect --period week
waymark reflect --period week --save
Reflections are app-only and source-grounded: they summarize counts, types, tags, and recent titles rather than inventing patterns. Saved reflections build a history you can list, compare, and trend over time.
8. Track a decision¶
waymark decision add "Adopt local embeddings?" \
--context "Hybrid search would help retrieval." --memory 1
waymark decision finalize 1 --choice "Yes, opt-in" --confidence 4
waymark decision outcome 1 --outcome "Retrieval noticeably improved."
Where to go next¶
- Importing your world — bring in Markdown, text, PDF, and Word documents.
- Daily loop — return tomorrow and know what needs attention.
- Local AI with Ollama — opt into local model structuring and semantic search.
- Backup & restore — keep your memory trail safe and portable.
- CLI reference — every command and option.