Evaluator path¶
Use this page when you want the shortest honest route through openclaw-mem.
If you have 5 minutes¶
Goal: prove the core mechanism on synthetic memory only.
git clone https://github.com/phenomenoner/openclaw-mem.git
cd openclaw-mem
uv sync --locked
uv run --python 3.13 --frozen -- \
python benchmarks/trust_policy_synthetic_proof.py --json
Expected result:
passed: true- one quarantined row selected by vanilla packing is excluded by trust-aware packing
- selected rows keep citation coverage
- the trust policy explains the exclusion reason
Read next: Trust-policy synthetic proof.
If you have 30 minutes¶
Goal: decide whether the sidecar-first adoption path fits your operator workflow.
- Run the trust-policy synthetic proof.
- Run the 60-second reality proof.
- Read Core vs Advanced Labs.
- Read Choose an install path.
- Skim Automation status so you know what is automatic, opt-in, partial, or roadmap.
Decision point:
- If you only need a short chat memory, stop here.
- If you need inspectable agent memory with citations and rollback posture, try the sidecar path.
- If you need live-turn recall orchestration, evaluate the optional mem engine after sidecar proof.
If you have one afternoon¶
Goal: evaluate whether openclaw-mem belongs in a real OpenClaw operator stack.
- Run the 5-minute proof and save the JSON receipt.
- Install the sidecar using Quickstart and Install modes.
- Ingest a small sanitized test memory file, not production memory.
- Run
search,timeline,get, andpackon 3–5 realistic questions. - Inspect citations and trace receipts for at least one expected include and one expected exclude.
- Only then consider optional mem-engine promotion.
Keep Advanced Labs out of the first evaluation unless your use case explicitly needs them.