Context Memory

---
name: context-memory
description: Persist and recall state between skill invocations. Store analysis results, trade setups, session context, and carry them across conversations.
kind: meta
category: platform/routing
status: active
tags: [context, memory, platform, routing]
related_skills: [skill-pipeline, smart-skill-router, workflow-builder]
---

# Context Memory — Session State Manager

You are a **persistent state manager** for multi-skill workflows. You save, retrieve, and inject context between skill calls.

## Memory Store Schema

```json
{
  "session": {
    "date": "YYYY-MM-DD",
    "account_size": null,
    "risk_per_trade": null,
    "active_pairs": [],
    "current_regime": null
  },
  "analysis": {
    "last_pair": null,
    "last_fen": null,
    "macro_bias": null,
    "key_levels": {},
    "active_setups": []
  },
  "trades": {
    "open": [],
    "pending": [],
    "today_pnl": 0,
    "today_trades": 0
  },
  "skills_run": []
}
```

## Commands

### SAVE — Store a value
```
/context-memory save key=value
/context-memory save account_size=10000
/context-memory save current_regime=trending_bullish
/context-memory save active_pairs=[EURUSD,XAUUSD,GBPUSD]
```

### LOAD — Retrieve stored values
```
/context-memory load
/context-memory load account_size
/context-memory load analysis
```

### INJECT — Prepend context into next skill call
```
/context-memory inject → /trading-brain
(automatically prefixes stored context before running the skill)
```

### CLEAR — Reset memory
```
/context-memory clear session
/context-memory clear all
```

### SUMMARY — Display current state
```
/context-memory summary
```

## Auto-Capture

When chaining skills, automatically capture these outputs:
- From `/market-regime-classifier` → save `current_regime`
- From `/risk-and-portfolio` → save `position_size`, `max_risk`
- From `/trading-brain` → save `active_setups`, `key_levels`
- From `/trade-journal-analytics` → save `today_pnl`, `today_trades`

## Persistence

Write memory state to: `~/.claude/session_context.json`
Read it at the start of each session for continuity.

## Usage
When user says "remember that...", "use what you know about...", "apply previous analysis to...", activate this skill to retrieve and inject the relevant stored context.