---
skill_id: long-term-memory-weaver
evaluation: free
price_usd_full: 8.99
challenge: https://lvlltd.com/api/pay?skill=long-term-memory-weaver
outline: https://lvlltd.com/skills/long-term-memory-weaver/outline.json
standards: x402, ERC-7857
---

# Free sample — Long-Term Memory Weaver

**Skill ID:** `long-term-memory-weaver`  
**Full unlock:** $8.99 USDC via x402 on Base  
**Challenge:** https://lvlltd.com/api/pay?skill=long-term-memory-weaver  
**Outline JSON:** https://lvlltd.com/skills/long-term-memory-weaver/outline.json

## Summary

Production long-term memory weaving: extract facts and decisions, attach provenance, link entities, and recall with citations so agents stop inventing history.

## When to use

- Agent forgets decisions across sessions
- Need durable memory with provenance
- Building multi-session marketplace agents

## Decision criteria (buy vs skip)

- You need this when: Agent forgets decisions across sessions
- You get capability: Define memory layers (working episodic semantic procedural)
- Budget fits: full unlock is $8.99 USDC on Base
- You can automate unlock via HTTP 402 without a browser
- You only need teaser/outline first — sealed src/ops stay locked until payment

## Outline (preview)

1. Define memory layers (working episodic semantic procedural)
2. Extract candidate memories with confidence
3. Scrub secrets and PII
4. Weave links between entities and decisions
5. Store with tenant isolation
6. Recall with citations only

## Capabilities (preview)

- Define memory layers (working episodic semantic procedural)
- Extract candidate memories with confidence
- Scrub secrets and PII
- Weave links between entities and decisions
- Store with tenant isolation
- Recall with citations only

## What you get after payment

- SKILL.md (teaser public)
- agent-install.json (public meta)
- src/index.js (stub public; full after unlock)
- Full sealed: src/, schemas/, ops/* (when present) after verified X-PAYMENT

## Agent buy path (zero human UI)

```
GET  https://lvlltd.com/skills/long-term-memory-weaver/outline.json
GET  https://lvlltd.com/api/pay?skill=long-term-memory-weaver   → HTTP 402
# transfer maxAmountRequired USDC to payTo on Base
POST https://lvlltd.com/api/pay
  X-PAYMENT: {"txHash":"0x…","skill":"long-term-memory-weaver"}
# response.sealed_pack.files = full pack
```

## Sample excerpt

# Long-Term Memory Weaver

**Price:** $8.99 / unlock
**Category:** Memory

## Summary

Production long-term memory weaving: extract facts and decisions, attach provenance, link entities, and recall with citations so agents stop inventing history.

---

---
name: long-term-memory-weaver
description: Weave durable long-term agent memory from sessions into entity episode and decision layers with provenance. Use when long-term memory weaver, durable agent memory, weave memory across sessions, agent memory architecture, or /long-term-memory-weaver.
---

# Long-Term Memory Weaver

Production long-term memory weaving: extract facts and decisions, attach provenance, link entities, and recall with citations so agents stop inventing history.

## Quick Reference
| Item | Value |
|------|-------|
| Category | Memory |
| Unlock | $8.99 USDC (x402 Base) |
| Triggers | long-term memory weaver, durable agent memory, weave memory, /long-term-memory-weaver |

## When to use
- Agent forgets decisions across sessions
- Need durable memory with provenance
- Building multi-session marketplace agents

## Core workflow
1. Inventory what must survive sessions
2. Extract candidates from session logs
3. Run secret/PII scrub gate
4. Dedupe and weave entity links
5. Write durable nodes with provenance
6. Verify recall with citation test

## Procedures (non-obvious)
### Node minimum fields
`id`, `type`, `content`, `created_at`, `source`, `confidence` (0-1), `tenant_id`, `tags[]`, `retain_until`.

### Weave rules
Prefer edges `decided`, `caused`, `mentions`, `supersedes` over free text. On conflict mark `conflict:true` and surface both.

### Recall
Never invent a memory. If miss say "no durable memory". Score = semantic + recency + confidence.

## Edge cases
- Conflicting decisions same day
- Delete-my-data requests need hard delete + tombstone

## Failure modes
- Storing secrets
- Unbounded growth without prune
- Recall without provenance

## Verification
- Write-recall roundtrip with citation
- Secret redaction test
- Tenant isolation negative test

## Artifact outputs
- `memory-weave.json`
- `entity-graph.md`
- `recall-report.md`

## Related skills
- context-budget-engineer
- agentic-rag-p

…(remainder sealed after x402 unlock)…