Files

T

Krystie dc5dc94d79 Initial commit: Adaptive Recall sidecar for UMP (Phase 5)

Multi-channel retrieval sidecar over Universal Memory Protocol:
- 3-channel RRF (UMP FTS5 + Qdrant vector + knowledge graph)
- ACT-R re-ranking (Anderson 1983) with access tracking
- Co-occurrence graph edges (Phase 6) for dense traversal
- Memory lifecycle decay (Phase 4) with per-kind confidence
- MCP shim routes recall through sidecar, falls back to canonical UMP

Architecture:
- src/server.js      HTTP sidecar on port 4380
- src/graph.js       2592-node / 111-edge graph from UMP (or +cooccur: 13k+)
- src/actr.js        A_i = -d*ln(age) + beta*log1p(freq) + epsilon*conf
- src/access_log.js  per-URN counter + last_accessed_at
- src/ump-recall-mcp.js  MCP shim (recall via sidecar, others passthrough)

Eval results (851-record UMP corpus):
- 2ch RRF over baseline: +50pp recall@10
- 3ch RRF (+graph): +60pp, 12 unique wins
- ACT-R re-rank: 4/20 #1 changes, 84% top-5 retention

Tests: 76/76 passing across graph (27), actr (27), access_log (28),
decay (20), mcp-shim (sidecar + fallback). Run with: npm test

Inspired by AIAppsAPI/adaptive-recall but built from scratch against
existing DNS2 infrastructure (UMP at :4317, Qdrant at :6333,
Ollama at :11434). No paid SaaS, MIT-licensed.

2026-07-12 19:24:47 -07:00

2.6 KiB

Raw Blame History

UMP memory.ump.json — actual schema (verified 2026-07-12)

File: /root/.openclaw/agents/main/workspace/state/ump-local/memory.ump.json Format: JSON array, 851 records Total records: 851 (not 846 — was 768 in Qdrant backfill, 846 in watcher's last run; real total is 851)

Top-level fields per record

{
  "ump": "0.1",                          // schema version (string)
  "id": "urn:ump:jzbhzerwqn23hfzre...",  // URN string (not "urn"!)
  "kind": "procedural",                  // see kinds below
  "body": {
    "subject": "...",                    // short title
    "text": "..."                        // full content
  },
  "scope": {
    "owner": "did:key:z6Mk...",          // actor DID
    "project": "openclaw/workspace",
    "visibility": "private",
    "tags": ["..."]
  },
  "time": {
    "created": "2026-06-06T21:49:57.806Z",  // ISO 8601
    "modified": "...",
    "observed": "...",
    "valid_from": "...",
    "valid_to": null
  },
  "lifecycle": {                         // NESTED, not flat!
    "status": "active",                  // see statuses below
    "confidence": 0.6                    // 0.0 - 1.0
  },
  "superseded_by": [],                   // array or null
  "provenance": {...},
  "integrity": {...}
}

Kinds (distribution)

kind	count
procedural	245
semantic	231
identity	152
episodic	152
note	42
working	29

(Total: 851)

Lifecycle statuses (distribution)

status	count	meaning
active	708	normal, queryable
candidate	53	pending promotion
archived	42	kept but deprioritized
tombstoned	48	deleted logically — skip in retrieval

Critical corrections vs spec

Field is id, NOT urn
lifecycle.status is nested in lifecycle.status, NOT flat status
lifecycle.confidence is nested, NOT flat
kind values include note (in addition to standard 5)
Status values: active, candidate, archived, tombstoned (not just active/archived)
No topic field — uses scope.project and scope.tags
No access_count or last_accessed_at fields yet (would need to be added)

Code patterns

import fs from 'node:fs';
const data = JSON.parse(fs.readFileSync(UMP_FILE, 'utf-8'));
for (const r of data) {
  const urn = r.id;                      // NOT r.urn
  const text = r.body?.text ?? r.body?.subject ?? '';
  const kind = r.kind;                   // procedural|note|semantic|working|identity|episodic
  const status = r.lifecycle?.status ?? 'active';
  const confidence = r.lifecycle?.confidence ?? 1.0;
  const createdAt = r.time?.created ?? null;
}

2.6 KiB Raw Blame History