Skip to content

Reality check & status

This page is intentionally factual.

Legend

  • DONE — works locally and deterministically (no network required unless stated)
  • PARTIAL — shipped, but optional/experimental and/or depends on external services
  • ROADMAP — planned; not shipped yet

Reality check (verifiable in ~60s)

1) Install deps (from a fresh clone)

uv sync --locked
uv run --python 3.13 --frozen -- python -m openclaw_mem --help

Note: this repo defines the openclaw-mem console entrypoint, so uv run openclaw-mem ... is supported after dependency install (uv sync --locked). The docs use uv run --python 3.13 --frozen -- python -m openclaw_mem ... as the most explicit source-checkout form.

2) Local DB smoke test (no OpenClaw required)

DB=/tmp/openclaw-mem-realitycheck.sqlite

uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json status

cat > /tmp/openclaw-mem-sample.jsonl <<'JSONL'
{"ts":"2026-02-12T08:10:00Z","kind":"tool","tool_name":"web_search","summary":"searched docs status markers","detail":{"query":"status markers DONE PARTIAL ROADMAP"}}
{"ts":"2026-02-12T08:11:00Z","kind":"note","summary":"Docs updated","detail":{"path":"README.md"}}
JSONL

uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json ingest --file /tmp/openclaw-mem-sample.jsonl
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json search "Docs" --limit 5
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json timeline 2 --window 2
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json get 2
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json optimize review --limit 200
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json optimize consolidation-review --limit 200
uv run --python 3.13 --frozen -- python -m openclaw_mem --db "$DB" --json optimize policy-loop --review-limit 200 --writeback-limit 100 --lifecycle-limit 100

Expected output shape (minimal):

  • status returns a JSON object containing at least:
  • db, count, min_ts, max_ts
  • embeddings.count and embeddings.models
  • ingest returns something like:
  • { "inserted": 2, "ids": [1,2] }
  • search returns a JSON array of compact rows containing at least: id, ts, kind, summary, plus snippet and score (it does not include detail_json).
  • timeline and get return JSON arrays of full rows containing at least: id, ts, kind, summary, and detail_json.
  • optimize review returns a recommendation-only report (openclaw-mem.optimize.review.v0) with signal counts and non-destructive recommendations, including signals.recent_use and staleness.protected_recent_use when recent pack-lifecycle evidence exists.
  • optimize consolidation-review returns a recommendation-only episodic maintenance report (openclaw-mem.optimize.consolidation-review.v0) with summary/archive/link candidates, source episode refs, recent-use archive protection, and receipt-first link evidence from lifecycle co-selection plus bounded lexical backfill (cold-start lexical fallback when lifecycle rows are absent).
  • optimize policy-loop returns a read-only rollout report (openclaw-mem.optimize.policy-loop.v0) with Stage B/C gate status, writeback linkage, and lifecycle-shadow evidence (no mutations).

3) Engine receipt debug smoke (local-only, no memory text)

node --experimental-transform-types tools/mem-engine-receipts-debug.mjs

Expected: - prints one synthetic recall lifecycle receipt (openclaw-mem-engine.recall.receipt.v1) - prints one synthetic autoCapture lifecycle receipt (openclaw-mem-engine.autoCapture.receipt.v1) - payload contains IDs/scores/counts only (no memory content text)

Status map (operator view)

Core (local, deterministic) — DONE

  • SQLite ledger + FTS5
  • JSON receipts (--json)
  • Progressive recall: ingest → search → timeline → get
  • Context pack command (pack) with fail-open baseline behavior
  • Deterministic graph-preflight pack integration: pack --use-graph=off|auto|on with traceable trigger/probe receipts
  • Pack decision surfaces (trust_policy, graph provenance_policy, policy_surface, lifecycle_shadow) with bounded trace receipts
  • Deterministic triage for automation (triage --mode heartbeat|cron-errors|tasks)

Quality layers — PARTIAL

  • Embeddings + vector search (embed, vsearch) — requires an API key
  • Hybrid retrieval (hybrid) + optional rerank providers — requires network/provider; still being evaluated
  • AI compression (summarize) — requires an API key
  • Graph query plane (graph topology-refresh, graph query ..., drift/provenance checks) — shipped foundation; deeper integrations still evolving
  • Recommendation-only optimization observers (optimize review, optimize consolidation-review) — shipped in the current release scope; now include recent-use-aware decay protection while still proposing only and never mutating stored memories
  • Dual-language fields (--text-en, EN embedding table) — shipped, still evolving
  • Episodic event capture/ingest/query lane — shipped foundation; operator flows still evolving

OpenClaw integration — PARTIAL

  • Auto-capture plugin (extensions/openclaw-mem) — captures tool results to JSONL
  • Backend-aware annotations (records backend + memory tool actions for observability)
  • Gateway-assisted semantic recall (Route A): index + semantic — depends on OpenClaw gateway + memory_search

Near-term roadmap — ROADMAP

  • Package/distribution ergonomics so openclaw-mem install/run flow is cleaner across uv sync / pip contexts
  • Graph roadmap depth: richer provenance wiring + higher-level operator queries on top of the shipped query plane
  • Topology seed automation (topology-seed) so curated graph truth is easier to bootstrap and maintain Topology seed automation (topology-seed) so curated graph truth is easier to bootstrap and maintain