Index of every page on this site: llms.txt
All skills
tradermonty avatar

/portfolio-manager

@f95e770
Run it yourself

Comprehensive portfolio analysis using Alpaca MCP Server integration to fetch holdings and positions, then analyze asset allocation, risk metrics, individual stock positions, diversification, and generate rebalancing recommendations. Use when user requests portfolio review, position analysis, risk assessment, performance evaluation, or rebalancing suggestions for their brokerage account.

Use this Skill: https://skilld.dev/gh/tradermonty/claude-trading-skills/portfolio-manager

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencescsv-holdings-review.md

≈2.1k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Offline holdings review

Run from the repository root with Python 3.9+; no API keys or third-party imports:

python3 skills/portfolio-manager/scripts/review_csv.py \
  --input skills/portfolio-manager/assets/fictional_holdings.csv \
  --as-of 2026-10-20T12:00:00Z --output-dir reports

For a standalone package, run python3 scripts/review_csv.py --input assets/fictional_holdings.csv --as-of 2026-10-20T12:00:00Z --output-dir reports from the extracted skill directory. The fixture contains fictional USD equities. Supply the actual source snapshot timestamp for your own input; the script does not verify prices, ownership, completeness or freshness. It never fetches quotes.

CSV input contract

UTF-8 CSV (optional BOM), one header row, comma separator, at most 10,000 holdings. Required headers: symbol,asset_type,quantity,direction,price,currency. Optional headers: cost_basis,account. Header order is arbitrary; unknown, missing and duplicate headers are errors. Every record must match the header width.

Field Contract
symbol Strip surrounding spaces, uppercase; [A-Z][A-Z0-9.-]{0,23}. Operator asserts this identifies the intended instrument.
asset_type Exactly equity; options, futures and other instruments are unsupported.
quantity Positive share count; fractional shares supported.
direction Exactly LONG or SHORT; quantities are always positive.
price Positive snapshot price per share, supplied by operator.
currency Exactly USD; currency conversion is unavailable.
cost_basis Optional nonnegative acquisition price per share; blank/missing means unknown. Excludes fees, dividends and tax adjustments.
account Optional ignored field, discarded before reporting. Review one account at a time; never use this column to infer account completeness.

Numbers use plain unsigned decimal notation, at most 32 characters and 12 fractional digits, and cannot exceed 1e12. Exponents, signs, whitespace, NaN, infinity and zero quantity/price are invalid. A zero cost basis is permitted. Duplicate normalized symbols, including separate tax lots or long/short rows, are rejected: consolidate them manually into the intended single holding first; do not net hedges merely to bypass the validator. Empty/header-only files fail; this tool does not establish that an account is empty.

Saved Alpaca positions input

A saved JSON list from the Alpaca /v2/positions endpoint can use the same limited analysis. Save the positions response locally through your existing read-only brokerage connection; this command does not connect to Alpaca:

python3 skills/portfolio-manager/scripts/review_csv.py \
  --input skills/portfolio-manager/assets/fictional_alpaca_positions.json \
  --input-format alpaca-json --as-of 2026-10-20T12:00:00Z --output-dir reports

The fictional JSON represents the same holdings as the fictional CSV. Only the following fields enter the common CSV validator/calculator:

Alpaca field Required mapping
symbol String, normalized and validated like CSV symbol
asset_class String, exactly us_equity, mapped to equity
qty String; strictly positive for side=long, strictly negative for side=short; its unsigned magnitude follows CSV numeric limits
side String, exactly long or short, mapped to LONG or SHORT
current_price String, positive USD per-share price with at most 12 decimal places, mapped to price
avg_entry_price Optional string, nonnegative per-share acquisition price with at most 12 decimal places, mapped to cost_basis; null, empty or missing means unavailable
currency If supplied, exactly USD; us_equity snapshots without this field are treated as USD only

See Alpaca's positions reference for native position fields. This adapter supports the restricted contract above, not every instrument or currency that the endpoint can return. Numbers must be JSON strings, not floats, so input precision is not silently altered. Contradictory quantity signs are errors. Input must be a nonempty list of at most 10,000 objects; duplicate normalized symbols are rejected just as for CSV.

All other fields, including account/asset IDs, names, provider totals, cost_basis (a position total rather than the required per-share price), weights, and provider P&L, are ignored. Missing per-share entry price stays unavailable; the adapter does not infer it from a position total. It recomputes all supported analysis through the same decimal calculator as CSV. Inputs and diagnostics do not prove that the saved response is current, complete or authentic. Supply its actual source timestamp; historical analysis with a current saved snapshot is not point-in-time data.

Shared limited output contract

The default --input-format csv retains the existing command, JSON schema and filenames. Both formats write portfolio_csv_review.json and portfolio_csv_review.md into the output directory, replacing previous reports with those names. The legacy filename does not identify the input source: read provenance.source. Use a new directory to retain history. Validation errors return exit 1, write a safe stderr diagnostic with row/field context, and create no reports. Argument errors return exit 2. I/O errors return exit 1; if writing fails midway, discard both report files and rerun. Existing reports are not proof that the latest invocation succeeded.

The structural JSON Schema is assets/holdings-review.schema.json. Runtime input validation enforces the numeric bounds and calculations described here. JSON schema_version: 1, report_type: holdings_allocation_only contains:

  • provenance: source: local_csv or alpaca_positions_snapshot, UTC as_of, freshness: operator_asserted. This is source-snapshot time supplied by the operator, not a fabricated fetch time.
  • gross_holdings_value: sum of quantity × price across all positions.
  • net_holdings_value: signed sum (shorts negative). Neither is account equity.
  • positions: sorted sanitized symbols, asset_type, direction, quantity, price, currency, nullable cost_basis, gross_market_value, net_market_value, gross_weight_pct and nullable unrealized_pnl. Decimal values are strings to preserve precision. Weights use gross holdings value, rounded to six decimal percentage places; rounded weights may not sum exactly to 100.
  • unrealized_pnl: quantity × (price − cost_basis), sign reversed for shorts. Unknown costs yield null, rendered unavailable. No aggregate P&L is emitted.
  • limitations: explicit unavailable analyses and integration boundaries.

Equivalent CSV and Alpaca snapshot holdings produce identical calculation fields; only source provenance and the Markdown source label differ. Both fictional fixtures yield gross value 80, signed net value 20, gross weights 62.5% and 37.5%, and per-position unrealized P&L of 12.5 and 10 USD. Cash, equity, margin/leverage, sector allocations, beta, historical returns, dividend health and trading recommendations are unavailable. Gross weights can look benign in a leveraged account: never use them as account exposure or as authorization to trade. Full live brokerage reports have a broader contract; this limited contract is not a substitute for that analysis.

Privacy and workflow boundary

No network, broker access, orders, memory edits or reconciliation gate. Input filenames/paths, account fields and IDs are excluded from script diagnostics and output. Reports still contain potentially sensitive holdings, quantities and cost data; keep inputs/reports local and do not commit or upload personal data. This is field minimization, not anonymization. CLI arguments/paths can remain in shell history. Never pass secrets in a command, CSV or saved JSON.

The portfolio-holdings-review workflow and Navigator's local-file route use this limited review. Navigator recommends the workflow without reading or validating your file; actual execution requires --input and an aware --as-of and may reject unsupported or malformed input. Its native replay uses fictional holdings and executes this CLI, not a brokerage request. It does not write a journal or trader-memory record. The broader core-portfolio-weekly workflow still requires broker/memory reconciliation and brokerage access. Do not bypass that gate or present a CSV allocation report as a completed full account review.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 alert3d5 checks · Risk SAFE
  • Gen Agent Trust Hub4d

    The skill is safe but has a minor vulnerability surface for indirect prompt injection through the processing of portfolio data files.

  • Socket3d

    No alerts

  • Snyk4d

    Risk: MEDIUM · 1 issue

  • Runlayer8mo

    5/11 files flagged

  • ZeroLeaks7mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at f95e770. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 5 days ago

README badge

README badge for tradermonty/claude-trading-skills/portfolio-manager