Studio MCP server reference
Every tool, prompt and resource the Last Vesper Studio MCP server offers, what each one does, and the limits that apply. For people connecting Claude, ChatGPT, Cursor or another MCP client.
Address, sign-in and access
The server lives at https://lastvesper.com/api/mcp and speaks the streamable HTTP transport. Apps that support sign-in (Claude, ChatGPT, Cursor) send you to a Last Vesper page where you pick one of your own books and choose read only or read and write. Developer tools can instead send a token from the Studio as an Authorization: Bearer header. The connection guide has the steps for each app.
AI agents are a Pro feature, and every new account starts on a free Pro trial. One connection sees one book: the one you picked, never anyone else's, and never the platform's own book.
Everything is paper trading. No tool places an order, reaches a broker, takes a strategy live or reveals a sealed rule, and no tool creates or revokes tokens.
Read tools
Available to every connection.
book_summary: One call for the whole book: strategies and read dates, unread inbox, edge monitor states and today's picks.
get_signals: Today's picks for your active paper strategies. Picks only: nothing is placed.
get_profile: Your Studio profile: capital, risk, holdings and banned tickers.
get_fundability: Which structures your capital can hold.
list_features: The rule features a strategy can use, with their history depth.
research_memory: What everyone's past tests say about each kind of idea: how crowded it is, which ideas have already failed, and the lessons.
list_strategies: Your registered strategies with their status and read dates.
get_performance: Paper results of your strategies.
holdings_fit: How your recorded positions fit your profile.
get_recommendations: Ranked strategies that fit your profile.
edge_monitor: Whether each strategy's recent results changed because the market paid less for selling options (regime) or because the strategy itself got worse (decay).
get_inbox: Your Studio inbox: picks, finished tests and read dates.
inbox_unread: Count of unread inbox items.
generate_strategies: Complete draft rules written from your profile and the shared library. Nothing is scored or registered.
get_calibration: How your real fills compare to paper.
list_tests: Your recent history tests.
get_test: One history test and its result. Poll this after run_test.
registered_ideas: The ideas you already have registered, so an agent does not register the same one twice.
strategy_library: The shared strategy library: tried ideas with their outcomes, including rejected ones.
list_public_strategies: Public strategies from other people and from the platform's own research, which you can copy and change.
get_handle: Your public author handle.
validate_rule: Check a rule and see what has been tried like it. Spends nothing.
plan_registration: What register_quick would register: read date, bar and notes. Registers nothing.
exposure_check: Factor exposure of a pasted positions CSV. Stores nothing.
Write tools
Need read and write access. History tests and AI drafts count against the same daily limits as the Studio itself.
test_and_wait: Run a history test and wait for it, up to about a minute. If it is still running, poll get_test with its id.
update_profile: Change your Studio profile. Fields left out are unchanged.
run_test: Queue a history test of a rule (kind evaluate scores it; coverage, regime, sweep, overlap and premium are the others). Counts against your daily test limit; poll get_test.
register_strategy: Register a rule to paper trading with its pre-registered read. Paper only.
register_quick: Register the plan from plan_registration to paper, private. Send confirm: true.
retire_strategy: Retire one of your strategies. Its record stays.
mark_inbox_read: Mark every inbox item read.
set_handle: Set or clear your public author handle.
import_holdings: Record positions in your own book from a broker CSV.
critique_rule: A written critique of a rule. Counts against your daily draft limit.
draft_from_text: Turn a plain description into a rule. Counts against your daily draft limit.
refine_rule: Change a rule by instruction. Counts against your daily draft limit.
rule_variations: Variations of a rule. Counts against your daily draft limit.
Prompts
design_strategy: Draft, check and test a strategy that fits your profile, then offer to register it.
review_book: How each paper strategy is doing, what the edge monitor says, and what needs attention.
explain_test: What one history test result means, and what to try next.
Resources
lastvesper://studio/rule-reference: Every field, feature, structure and management rule a Studio strategy can use.
lastvesper://studio/example-rule: A complete, valid rule to start from.
The rule format is generated from the same validator the Studio uses, so an agent that follows it writes rules the Studio accepts.
Limits and safety
Each token or connected app can make 60 requests a minute. Every write is recorded in your book's audit trail, and the token panel shows how many calls each token and app made in the last 7 days.
Disconnect an app or revoke a token at any time under Signals and AI agents in the Studio. A connected app's access lasts an hour at a time and renews itself until you disconnect it.
Historical replays are not a forecast. Options can lose more than the premium collected. This is education, not a recommendation to trade anything.