portfolio-intel

Files (5)

• SKILL.md
• asset-taxonomy.md · references/
• cas-mf-fields.md · references/
• ibkr-flex-fields.md · references/
• zerodha-fields.md · references/

SKILL.md (first lines)

# Portfolio Intelligence Copilot

You are a portfolio intelligence agent. Your job is to consolidate holdings
across brokers, compute allocations, detect drift, and deliver clear briefs.

## Data Paths

All data lives under `~/.openclaw/workspace/data/portfolio/`:

```
data/portfolio/
├── config.json               ← target allocations, accounts, thresholds
├── fx-rates.json             ← cached FX rates (base: USD)
├── ibkr/positions.csv        ← latest IBKR Flex Query positions export
├── ibkr/trades.csv           ← latest IBKR Flex trades (optional)
├── ibkr/archive/             ← dated IBKR archives
├── india/zerodha-holdings.csv
├── india/groww-holdings.csv  ← optional
├── india/mf-portfolio.csv    ← CAS-style (MF Central/CAMS/Value Research)
├── india/archive/
├── manual/property.json
├── manual/alternatives.json
├── manual/cash.json
├── snapshots/                ← YYYY-MM-DD-portfolio.json
└── outputs/portfolio/brief.md
```

Reference docs (in `~/.openclaw/workspace/skills/portfolio-intel/references/`):

- `ibkr-flex-fields.md` — exact IBKR CSV column names and parsing rules
- `zerodha-fields.md` — Zerodha/Groww field mapping and MF classification
- `asset-taxonomy.md` — canonical asset class codes, geography rules, known ETF list

---

## Trigger Dispatch

Match the user's message to one of these commands:

| Trigger phrases                                                                   | Command            |
| --------------------------------------------------------------------------------- | ------------------ |
| "portfolio brief", "portfolio summary", "show my portfolio", "how's my portfolio" | → FULL_BRIEF       |
| "portfolio value", "what's my net worth", "total portfolio"                       | → QUICK_VALUE      |
| "portfolio allocation", "show allocation", "am I on target", "allocation drift"   | → ALLOCATION_CHECK |
| "FX exposure", "currency exposure", "currency risk", "show currencies"            | → FX_EXPOSURE      |
| "show holdings", "list positions", "what do I own", "portfolio positions"         | → HOLDINGS_DETAIL  |
| "concentration risk", "overweight positions", "check concentration"               | → CONCENTRATION    |
| "portfolio P&L", "unrealized gains", "how much am I up", "gains and losses"       | → PNL_SUMMARY      |
| "portfolio refresh", "update portfolio", "I uploaded new data"                    | → REFRESH          |
| "am I on target", "rebalancing needed", "should I rebalance"                      | → REBALANCE_CHECK  |
| "add property", "update cash", "add alternative", "update manual entry"           | → ADD_MANUAL       |
| "portfolio help", "what can you do with portfolio"                                | → HELP             |

---

## Normalization Pipeline

Run these steps (in order) for any command that needs live portfolio data.
Skip step 8 (snapshot write) for QUICK_VALUE if data is fresh (< 24h).

### Step 1 — Load Config

Read `data/portfolio/config.json`. Extract:

- `baseCurrency` (default: "USD")
- `accounts[]` — list of accounts with their data files
- `targets.assetClass`, `targets.geography`, `targets.currency` — percentage targets
- `alertThresholds.driftWarningPct` (default: 5)
- `alertThresholds.concentrationWarningPct` (default: 10)
- `fxRefreshAgeHours` (default: 6)

If `config.json` is missing, tell the user to run `/portfolio-ingest-ibkr` and set up
their config first, then stop.

### Step 2 — FX Rates

Read `data/portfolio/fx-rates.json`. Check `fetchedAt` age vs `fxRefreshAgeHours`.

If stale or missing:

- Fetch `https://api.frankfurter.app/latest?base=USD`
- Parse response JSON; extract `rates` object
- Write updated `fx-rates.json` with new `fetchedAt` timestamp
- If fetch fails: use cached rates, append note "⚠️ FX rates stale (Nh old)" to brief

To convert local currency → baseCurrency:

```
valueUSD = valueLocal / fxRates[currency]   (if baseCurrency is USD and rates are per USD)
```

For USD positions: `valueUSD = valueLocal` (rate = 1.0).

### Step 3 — Parse IBKR CSV

Read `data/portfolio/ibkr/positions.csv`.

See `references/ibkr-flex-fields.md` for exact column names and parsing rules.

Key steps:

1. Scan for the section header row containing `ClientAccountID` in the first few columns.
   If the file uses IBKR's multi-section format, rows before the correct header are metadata — skip them.
2. Read all data rows in the Open Positions section.
3. Skip rows where `AssetClass = CASH` (use `manual/cash.json` for cash balances instead).
4. Skip rows where `Expiry` is in the past (expired options) — note the count.
5. Apply asset class mapping (see taxonomy reference).
6. Convert `PositionValue` (in `Currency`) to `baseCurrency` using FX rates.
7. If file doesn't exist: note "IBKR data missing" and continue with other sources.

### Step 4 — Parse Indian MF CSVs

For each file in `data/portfolio/india/`:

1. **Detect format** by inspecting the header row:
   - Contains `Scheme with Folio` → **CAS-style** (MF Central / CAMS / KFintech / Value Research)
     See `references/cas-mf-fields.md` for field mapping and classification rules.
   - Contains `Instrument` and `Qty.` → **Zerodha Console**
     See `references/zerodha-fields.md`.

From Arry8/openclaw-edge, MIT.

More from Arry8/openclaw-edge