Skip to main content

Usage

After installation and configuration, the PortDive MCP server is available as a tool in your AI assistant. Here's how to use it in your daily workflow.

Available tools

The MCP server provides 47 tools across 8 categories:

Market data

ToolDescription
list_available_tickersAll available tickers with exchange/timeframe availability and data freshness
get_ohlcv_with_indicatorsOHLCV candles with technical indicators — multi-timeframe, compact columnar format

Multi-timeframe analysis (MTA)

ToolDescription
get_mta_snapshotCurrent MTA snapshot: composite score/grade, signal breakdown, position sizing with ATR stops
list_mta_snapshotsMTA snapshot history — track how a ticker's MTA grade changes over time

Investment theses

ToolDescription
list_thesesList theses — own, official, or filter by status
get_thesisFull thesis with baskets, ticker groups, and clone attribution
create_thesisCreate a new thesis with inline basket and ticker definitions
update_thesisUpdate a thesis — only changed fields are overwritten
delete_thesisPermanently delete a thesis
get_thesis_signalsAggregated trading signals for a thesis or basket
get_thesis_performanceSynthetic equal-weight portfolio performance since creation
get_thesis_historyVersion history — every mutation is automatically archived
list_thesis_templatesStarter templates for new theses (de/en)

Basket management

ToolDescription
add_basketAdd a new basket (ticker group) to a thesis
update_basketUpdate basket metadata (description, entry/exit rules)
manage_basket_itemsAdd or remove tickers in a basket

Skills & prompts

ToolDescription
list_skillsBrowse skills — official, community, or your own
get_skillGet skill details: body, parameters, tags, resources, playbook DAG
run_skillRun a skill — assembles the prompt with parameters and referenced tools
create_skillCreate a custom skill — reusable prompt templates with {{param}} placeholders
update_skillUpdate a skill — automatically archives the previous version
delete_skillDelete your own skill
clone_skillClone a public skill into your workspace
get_skill_resourceGet a skill resource (reference docs, scripts)
get_skill_historySkill version history
attach_skill_to_thesisAttach a skill to a thesis — appears as a recommended workflow
list_thesis_skillsList skills attached to a thesis
validate_playbook_dagValidate a playbook DAG structure before saving
add_skill_resourceAdd a resource to a skill
delete_skill_resourceRemove a resource from a skill
upload_skill_assetUpload a file asset to a skill
delete_skill_assetDelete a skill file asset

Analysis & health

ToolDescription
get_thesis_healthComprehensive health check: signals, performance, version history, skills, data freshness
suggest_thesis_structureContext for AI-assisted thesis structuring with all available tickers

Alpha log

ToolDescription
add_alpha_log_entryLog an analysis result, decision, or snapshot
list_alpha_log_entriesTimeline entries filtered by thesis, ticker, or skill
get_alpha_log_entryFull alpha log entry with markdown content
upload_alpha_log_attachmentAttach a file to an alpha log entry
generate_alpha_log_upload_urlGenerate a presigned upload URL for file attachments
delete_alpha_log_attachmentDelete an alpha log attachment

AI prompts

In addition to tools, 7 guided AI prompts are available that walk Claude through structured analysis workflows:

PromptDescription
draft-thesisBuild a structured thesis from an idea — baskets, tickers, allocation
thesis-health-checkComprehensive health check: completeness, signals, diversification
contrarian-challengeDevil's advocate: bear cases, hidden risks, correlation traps
signals-briefingSynthesize current market signals across a thesis
rebalance-analysisAnalyze portfolio drift and rebalancing opportunities
ticker-due-diligenceDeep-dive on a single ticker within thesis context
basket-entry-exit-rulesGenerate markdown entry/exit rules for a basket

Creating investment theses

Thematic thesis: "AI Picks & Shovels"

Create a thesis around AI infrastructure — the companies selling "picks and shovels" to the AI gold rush:

Create a new investment thesis "AI Picks & Shovels" via PortDive.

Gist: This thesis invests in companies providing the infrastructure
for the AI boom — compute capacity, semiconductors, cloud platforms,
and data infrastructure.

Create two baskets:
1. "GPU & Compute" with NVDA (NASDAQ), AMD (NASDAQ), NBIS (NASDAQ)
2. "Cloud & Data" with AMZN (NASDAQ), MSFT (NASDAQ), GOOG (NASDAQ)

Benchmark thesis: "Mag 7"

Create a thesis "Magnificent 7" via PortDive.

Gist: The seven largest US tech companies as a benchmark portfolio.

One basket "Core" with: AAPL, MSFT, AMZN, GOOG, META, NVDA, TSLA
— all on NASDAQ.

Analyzing existing theses

Overview of your portfolios:

List all my investment theses via PortDive.
Show me the current performance and aggregated signals
with grade and verdict for each thesis.

In-depth health check:

Run a health check on my "AI Picks & Shovels" thesis via PortDive.
Check signal quality, data freshness, and diversification.

Devil's advocate against a thesis:

Use the PortDive "contrarian-challenge" prompt for my
"AI Picks & Shovels" thesis. Show me bear cases,
correlation risks, and scenarios where the thesis fails.

Applying official skills

Skills are reusable analysis workflows you can apply to any ticker. Official skills are available in the skill library.

Fundamental Valuation with NBIS

Show me the official skills via PortDive.
Apply the "Fundamental Valuation" skill to the ticker NBIS.

Claude uses list_skills (scope: official), finds the skill, and runs it with run_skill. The skill assembles a structured analysis prompt that incorporates the available OHLCV data and indicators for NBIS.

Macro Risk Overlay with SPY

Find the official PortDive skill "Macro Risk Overlay" and
apply it to SPY. Assess the current macro risks
for the S&P 500.

Attaching skills to theses

Assign recurring analysis workflows directly to your thesis:

Attach the "Fundamental Valuation" skill to my
"AI Picks & Shovels" thesis. Then run it for all tickers
in the "GPU & Compute" basket.

Attached skills appear as recommended workflows the next time you analyze the thesis.

Creating custom skills

Create a new PortDive skill "Weekly Thesis Review".

Body:
Run a weekly review for the thesis {{thesis_name}}:
1. Get current signals and performance
2. Compare with last week
3. Identify action items

Parameters: thesis_name (string, required)
Tags: review, weekly, thesis

Tips

  • Be specific: Name the ticker and timeframe when querying specific data
  • Combine tools: Claude can use multiple MCP tools in a single request to build a complete picture
  • Use prompts: The 7 guided AI prompts (draft-thesis, thesis-health-check, etc.) offer structured workflows for complex analyses
  • Alpha log: Record key findings with add_alpha_log_entry — they're available as context in future analyses
  • Markdown everywhere: Text fields like gist, description, and entry/exit rules support markdown formatting

Troubleshooting

MCP server not recognized

  1. Check that claude_desktop_config.json is properly formatted (valid JSON)
  2. Make sure the path /usr/local/bin/portdive-mcp exists
  3. Fully restart Claude Desktop (⌘ + Q, then reopen)

Authentication errors

  1. Check that your token is valid (not revoked)
  2. Create a new token in the PortDive settings if needed
  3. Make sure your subscription is active (INTELLIGENCE or PREMIUM)

Connection issues

  1. Check your internet connection
  2. Test the health check: portdive-mcp --health in the terminal
  3. The MCP server connects to https://api.portdive.app automatically — no manual configuration needed