/portfolio-review

name: portfolio-review
description: Composite skill that chains market-regime, sector-rotation-signal, risk-report, earnings-blackout, macro-event-calendar, corporate-actions-scanner, and portfolio-rebalancer into one call. Turns the manual 6-command portfolio-review workflow into a single invocation. Emits a headline summary that pulls the most decision-relevant fact from each section plus the full per-section detail below. Use when the operator asks "review my portfolio" or "run the full workflow on my book."

portfolio-review

You hand over positions + book value. The skill runs the full 7-tool review chain in the right order, threads the shared context (tickers, weights, lookback windows), skips ETFs on the equities-only tools (earnings, 8-K), and stitches the outputs into a single briefing.

The headline block is the read-first summary: regime, rotation theme, next earnings, next macro event, top corporate action, and the rebalance verdict in six lines. Full per-section detail follows.

When to invoke

• The operator says "review my portfolio", "run the full workflow",

"portfolio check", "what should I do with my book"

• Weekly / monthly portfolio hygiene
• After a material market move to sanity-check the book
• Before a large trade to see whether the new sizing survives the

variance-share cap

What you need

• `MASSIVE_API_KEY` (Stocks Starter minimum). One skill in the chain

(corporate-actions-scanner) also hits SEC EDGAR, which is free.

• Positions in the shape `TICKER=WEIGHT,TICKER=WEIGHT,...` where

weights sum to ~1.0.

• Book value in dollars for the rebalancer's trade-ticket sizing.

What you get back

**Layer 1 canonical JSON** with a `sections` map keyed by the sub- skill name, each holding the full sub-payload. The `headline` block distills one fact from each. `errors` array captures per-section failures without aborting the run.

**Layer 2 rendered briefing**. Header, headline block (7 lines), then each section rendered by its own `render()` helper under a titled divider. See [`references/rendering.md`](./references/rendering.md).

How it works

1. Parse positions, split into equities vs ETFs (ETFs skipped for earnings + 8-K scanners). 2. Run the 7 sub-skills in sequence, sharing a single MassiveClient for connection reuse: 1. `market-regime` 2. `sector-rotation-signal` 3. `risk-report` 4. `earnings-blackout` (equities only) 5. `macro-event-calendar` 6. `corporate-actions-scanner` (equities only) 7. `portfolio-rebalancer` (skippable via `include_rebalance=False`) 3. Build the headline by pulling the top-fact from each sub-payload. 4. Emit the composite payload with per-section detail and errors.

Foundations used

• All 7 sub-skills. This is a pure composition — no new data pulls

beyond what the sub-skills already fetch.

Endpoints used

• Aggregate of every sub-skill's endpoints. Chain sharing a

MassiveClient means daily-aggs fetches are cached per ticker across sub-skills where the cache applies.

Doesn't handle (yet)

• **Fixed income context.** No rates/credit read yet (waiting on

fixed-income-context, Part 3 #4).

• **Historical analogs.** Not included in the default chain because

regime-conditional forecasting is thesis-driven, not portfolio review. Callers who want it can invoke historical-analog-finder directly.

• **Options context.** Neither options-flow nor options-structure-

analyzer are part of the default chain; they're view-dependent, not review-dependent.

• **No changed-since-last-run diff.** Every review is stateless. A

future version could take a prior review payload and highlight what moved.