Cross-reference and synthesise daily memory files into patterns, insights, and carry-forward items. Use when asked to review memory, generate weekly synthesis, find forgotten tasks, detect decision patterns, or audit what was carried forward vs dropped. Also triggers on "what did I decide about X", "what's still pending", "weekly review", "memory synthesis", "what patterns do you see".
Scanned 9/2/2026
Install to Claude Code
npx -y skills add majiayu000/claude-skill-registry --skill memory-archivist --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Memory Archivist?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/majiayu000-memory-archivist-claude-skill-registry)More formats (shields.io, HTML) on the badges page.
---
name: memory-archivist
description: Cross-reference and synthesise daily memory files into patterns, insights, and carry-forward items. Use when asked to review memory, generate weekly synthesis, find forgotten tasks, detect decision patterns, or audit what was carried forward vs dropped. Also triggers on "what did I decide about X", "what's still pending", "weekly review", "memory synthesis", "what patterns do you see".
---
# Memory Archivist
Memory is sacred. This skill turns raw daily journals into structured wisdom.
## What This Skill Does
1. **Cross-reference** — Read all `memory/*.md` files and find connections between days
2. **Pattern detection** — Surface recurring themes, repeated decisions, persistent blockers
3. **Carry-forward audit** — Find items marked "carry forward" that were never resolved
4. **Weekly synthesis** — Generate a weekly summary from daily files
5. **Scar-to-decision mapping** — Track how USER.md scars influence actual decisions over time
## Trigger Patterns
- "review my memory", "memory synthesis", "weekly review"
- "what's still pending", "what did I forget"
- "what patterns do you see", "recurring themes"
- "what did I decide about [topic]"
- "carry forward audit", "dropped items"
- "synthesise this week"
## On Trigger — Execute in Order
### Step 1: Gather Memory Files
```bash
# List all memory files sorted by date
ls -1 ~/.openclaw/workspace/memory/*.md 2>/dev/null | sort
```
Read the most recent 7 files (or all if fewer than 7 exist). Use the `read` tool, not exec.
### Step 2: Extract Structured Data
From each daily file, extract:
| Field | Section to scan |
|-------|----------------|
| **Decisions** | "Key Decisions Made Today" |
| **Carry-forward** | "Carry Forward" |
| **888_HOLD** | "888_HOLD Items Pending" |
| **Floors active** | "Floors Active This Session" |
| **Context** | "Session Context" |
### Step 3: Cross-Reference
For each carry-forward item, check if it appears as resolved in any later day's decisions.
Build a table:
```markdown
| Carry-Forward Item | First Mentioned | Resolved? | Resolution Date | Days Open |
```
Flag any item open longer than 3 days as ⚠️ STALE.
### Step 4: Detect Patterns
Scan across all files for:
- **Repeated topics** — same subject appearing in 3+ daily files
- **Recurring blockers** — carry-forward items that keep reappearing
- **Floor patterns** — which floors are consistently active vs rarely active
- **Decision velocity** — how many decisions per day (trending up/down?)
### Step 5: Generate Synthesis
Output format depends on the request:
**Weekly synthesis** → Write to `memory/weekly-YYYY-WXX.md`:
```markdown
# Weekly Synthesis — YYYY-WXX (arifOS_bot)
## Period
YYYY-MM-DD to YYYY-MM-DD
## Key Decisions This Week
1. ...
## Carry-Forward Audit
| Item | Status | Days Open |
|------|--------|-----------|
## Patterns Detected
- ...
## 888_HOLD Items (Active)
- ...
## Floor Activity
F1: X sessions | F2: X sessions | ...
## Recommendation
<one-line suggestion for next week based on patterns>
```
**Ad-hoc query** ("what did I decide about X") → Search all memory files for the topic, return chronological trail of mentions with dates.
**Carry-forward audit** → Return the open items table only.
### Step 6: Seal to VAULT999
After generating synthesis, seal via arifOS MCP:
```bash
arifos seal
```
Or via HTTP:
```bash
curl -sf -X POST http://arifosmcp_server:8080/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"seal_vault","arguments":{"session_id":"memory-archivist","context":"weekly synthesis sealed"}}}'
```
This commits the synthesis to the immutable VAULT999 ledger. Memory is sacred — sealed memory is permanent.
### Step 7: Log the Synthesis
Append to `logs/audit.jsonl`:
```json
{"ts":"<ISO>","event":"memory_synthesis","source":"memory-archivist","files_scanned":<n>,"patterns_found":<n>,"stale_items":<n>,"sealed":true,"agent":"arifOS_bot"}
```
## Scheduling (Cron)
Run the archivist nightly at 00:30 MYT (16:30 UTC), after the git-sync backup at 00:00 MYT:
```bash
# Add to host crontab (not container — container restarts clear cron)
30 16 * * * cd /opt/arifos/data/openclaw/workspace && bash skills/memory-archivist/scripts/scan-carry-forward.sh memory/ >> logs/archivist.log 2>&1
```
For the full agent-driven synthesis (weekly), trigger manually or via HEARTBEAT.md on every 8th heartbeat (~4 hours):
- "Run memory archivist" or "weekly review"
## Constraints
- **Read-only on memory files.** Never edit existing `memory/*.md` files. Only create new synthesis files.
- **F1 Amanah.** All synthesis is additive. Nothing is deleted or overwritten.
- **F7 Humility.** If a pattern has low confidence (appears only twice), mark it as tentative.
- **F4 Clarity.** Output must reduce confusion. Tables over prose. One insight per line.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!